WHITE PAPER 
 
 
 
 
 
	
  
Shifting	
  the	
  goalposts—from	
  high	
  
impact	
  journals	
  to	
  high	
  impact	
  data	
  
Anja	
  Gassner1,	
  Luz	
  Marina	
  Alvare2,	
  Zoumana	
  Bamba3,	
  Douglas	
  Beare4,	
  Marichu	
  
Bernardo5,	
  Chandrashekhar	
  Biradar6,	
  Martin	
  van	
  Brakel7,	
  Robert	
  Chapman8,	
  Guntuku	
  
Dileepkumar9,	
  Ibnou	
  Dieng10,	
  Sufiet	
  Erlita11,	
  Richard	
  Fulss12,	
  Jane	
  Poole13,	
  Mrigesh	
  
Kshatriya11,	
  Guvener	
  Selim14	
  Reinhard	
  Simon14,	
  Kai	
  Sonder12,	
  Nilam	
  Prasai2,	
  Maria	
  
Garruccio8,	
  Simone	
  Staiger	
  Rivas14,	
  Maya	
  Rajasekharan14	
  ,	
  Chukka	
  Srinivasa	
  Rao9	
  	
  
	
  
	
  
	
  
	
  
1	
  World	
  Agroforestry	
  Centre	
  (ICRAF),	
  United	
  Nations	
  Avenue,	
  Gigiri,	
  PO	
  Box	
  30677	
  Nairobi	
  00100	
  Kenya	
  
2	
  International	
  Food	
  Policy	
  Research	
  Institute	
  (IFPRI),	
  2033	
  K	
  St,	
  NW,	
  Washington,	
  DC	
  20006-­‐1002	
  USA	
  
3	
  International	
  Institute	
  of	
  Tropical	
  Agriculture	
  (IITA),	
  HQ-­‐PMB	
  5320,	
  Ibadan,	
  Oyo	
  State	
  Nigeria	
  
4	
  WorldFish,	
  Jalan	
  Batu	
  Maung,	
  Batu	
  Maung,	
  11960	
  Bayan	
  Lepas,	
  Penang,	
  Malaysia,	
  	
  PO	
  Box	
  500	
  GPO,	
  10670	
  Penang,	
  
Malaysia	
  
5	
  International	
  Rice	
  Research	
  Institute	
  (IRRI),	
  DAPO	
  Box	
  7777	
  Metro	
  Manila	
  1301,	
  Los	
  Baños,	
  Philippines	
  
6	
  International	
  Center	
  for	
  Agricultural	
  Research	
  in	
  the	
  Dry	
  Areas	
  (ICARDA,	
  "Dalia	
  Building	
  2nd	
  Floor,	
  Bashir	
  El	
  Kassar	
  
Street,	
  Verdun,	
  Beirut,	
  Lebanon	
  1108-­‐2010	
  P.O.	
  Box	
  114/5055	
  Beirut,	
  	
  Lebanon"	
  
7	
  International	
  Water	
  Management	
  Institute	
  (IWMI),	
  P.	
  O.	
  Box	
  2075,	
  Colombo,	
  Sri	
  Lanka,	
  127,	
  Sunil	
  Mawatha,	
  Pelawatte,	
  
Battaramulla,	
  Sri	
  Lanka	
  
8	
  Bioversity	
  International,	
  "HQ-­‐	
  Via	
  dei	
  Tre	
  Denari	
  472/a	
  00057	
  Maccarese	
  (Fiumicino)	
  Rome,	
  Italy"	
  
9	
  International	
  Crops	
  Research	
  Institute	
  for	
  the	
  Semi-­‐Arid	
  Tropics	
  (ICRISAT),	
  Patancheru	
  502	
  324	
  Andhra	
  Pradesh,	
  India	
  
10	
  Africa	
  Rice	
  Center	
  (AfricaRice)	
  01	
  B.P.	
  2031,	
  Cotonou,	
  Benin	
  
11	
  Center	
  for	
  International	
  Forestry	
  Research	
  (CIFOR),	
  HQ-­‐	
  Jalan	
  CIFOR,	
  Situ	
  Gede	
  
Bogor	
  (Barat)	
  16115,	
  Indonesia,	
  Mailing-­‐P.O.	
  Box	
  0113	
  BOCBD	
  Bogor	
  16000,	
  Indonesia"	
  
12	
  International	
  Maize	
  and	
  Wheat	
  Improvement	
  Center	
  (CIMMYT),	
  Km.	
  45,	
  Carretera,	
  México-­‐Veracruz,	
  El	
  Batán,	
  Texcoco	
  
CP	
  56130,	
  Edo.	
  de	
  México,	
  Apdo.	
  Postal	
  6-­‐641,	
  06600	
  Mexico,	
  D.F.,	
  Mexico	
  
13	
  International	
  Livestock	
  Research	
  Institute	
  (ILRI),	
  HQ-­‐PO	
  Box	
  30709,	
  Nairobi,	
  00100	
  ,	
  Old	
  Naivasha	
  Road,	
  ,	
  Nairobi,	
  
Kenya	
  
14	
  International	
  Potato	
  Center	
  (CIP),	
  Avenida	
  La	
  Molina	
  1895,	
  La	
  Molina,	
  Apartado	
  Postal	
  1558,	
  Lima,	
  Peru	
  
15International	
  Center	
  for	
  Tropical	
  Agriculture	
  (CIAT),	
  Km	
  17,	
  Recta	
  Cali-­‐Palmira	
  
Apartado	
  Aéreo	
  6713,	
  Cali,	
  Colombia	
  
	
   	
  
	
  
Contents	
  
	
  
Introduction	
  ....................................................................................	
  3	
  
Types	
  of	
  research	
  data	
  ....................................................................	
  4	
  
Research	
  data	
  infrastructure	
  across	
  the	
  centers	
  ............................	
  6	
  
Success	
  stories	
  ..............................................................................	
  15	
  
Barriers	
  to	
  mainstreaming	
  data	
  sharing	
  .......................................	
  18	
  
Recommendations	
  ........................................................................	
  21	
  
Key	
  references	
  ..............................................................................	
  24	
  
	
   	
  
PAGE	
  -­‐	
  2	
  -­‐	
  
Introduction	
  
CGIAR	
   is	
  a	
  global	
  agriculture	
  research	
  partnership.	
   Its	
  science	
   is	
  carried	
  out	
  by	
  
the	
   15	
   research	
   centers	
   who	
   are	
   members	
   of	
   the	
   CGIAR	
   Consortium	
   in	
  
collaboration	
   with	
   hundreds	
   of	
   partner	
   organizations.	
   Each	
   member	
   of	
   the	
  
CGIAR	
  Consortium	
  has	
   the	
  mandate	
   to	
   contribute	
   to	
   the	
   eradication	
   of	
   hunger	
  
and	
   poverty	
   at	
   the	
   global	
   level	
   by	
   advancing	
   research	
   in	
   development.	
   Being	
  
funded	
  mainly	
   through	
  public	
   funds	
   there	
   is	
  an	
  obligation	
   to	
  extract	
  maximum	
  
public	
   good	
   value	
   from	
   the	
   research	
   data	
   the	
   individual	
   centers	
   are	
   collecting.	
  
With	
   the	
   exception	
   of	
   a	
   few	
   designated	
   projects,	
   in	
   the	
   past,	
   data	
   itself	
   was	
  
merely	
   seen	
  as	
  a	
  means	
   to	
  an	
  end,	
   synthesized	
   to	
  produce	
   selected	
  knowledge	
  
products	
  such	
  as	
  publications,	
  technical	
  manuals	
  or	
  policy	
  briefs.	
  
Changing	
   financial	
   realities	
   and	
   the	
   need	
   to	
   reach	
   our	
   aid	
   targets	
   with	
   less	
  
available	
   resources	
   is	
   leading	
   to	
   a	
   shift	
   in	
   thinking	
   for	
   new	
   ways	
   to	
   make	
  
scientific	
  research,	
  and	
  research	
  data,	
  more	
  available,	
  reusable	
  and	
  reproducible.	
  
Sufficiently	
   preserved	
   and	
   replicable	
   data	
   are	
   more	
   absolute	
   than	
  
contemporaneously	
  drawn	
  conclusions	
  and,	
  if	
  they	
  are	
  collected	
  to	
  address	
  one	
  
scientific	
   question,	
   can	
   later	
   be	
   applied	
   for	
   the	
   solution	
   of	
   entirely	
   different	
  
problems	
  (Altman	
  et	
  al.,	
  2006).	
  
A	
   growing	
   number	
   of	
   donors	
   now	
   require	
   data	
   sharing	
   requirements	
   in	
   their	
  
standard	
  grant	
  contracts	
  and	
  open	
  access	
   to	
  be	
  applied	
   to	
   the	
  research	
  data	
  of	
  
the	
  work	
  they	
  fund.	
  	
  DFID	
  and	
  the	
  European	
  Commission	
  are	
  among	
  those	
  who	
  
are	
   now	
   adopting	
   this	
   policy.	
   As	
   one	
   of	
   the	
   first	
   custodians	
   of	
   development	
  
related	
  data,	
   the	
  World	
  Bank,	
   launched	
   an	
   open	
  data	
   initiative	
   in	
   2012.	
   CGIAR	
  
itself	
   followed	
   this	
  example	
  and	
  approved	
   in	
  March	
  2012	
  a	
  policy	
   that	
  clarifies	
  
CGIAR	
   research	
   results,	
   including	
   data,	
   are	
   openly	
   available	
   and	
   accessible	
   by	
  
default	
   –	
   and	
   that	
   any	
   deviation	
   has	
   to	
   be	
   justified.	
   	
   Open	
   data	
   is	
   not	
   a	
   new	
  
concept	
  for	
  CGIAR,	
  but	
  in	
  the	
  past	
  has	
  largely	
  remained	
  linked	
  to	
  specific	
  projects	
  
or	
  core	
  programs	
  that	
  are	
  designed	
  to	
  collect	
  and	
  share	
  data	
  widely.	
  	
  
In	
  keeping	
  with	
   the	
   changes	
   to	
   the	
  amount,	
  management	
  and	
  analyses	
  of	
  data,	
  
the	
   role	
   of	
   scientists	
   working	
   in	
   development,	
   such	
   as	
   those	
   at	
   the	
   CGIAR	
  
Centers,	
   is	
   changing.	
   Previously,	
   grant	
   awards,	
   and	
   performance	
   evaluations	
  
focused	
   strongly	
  on	
  publication	
   in	
  high-­‐impact	
   journals.	
   Fortunately,	
   there	
   is	
   a	
  
rediscovery	
  that	
  the	
  real	
  currency	
  of	
  research	
  and	
  scientific	
  knowledge	
  –the	
  data	
  
methods	
   and	
   ideas-­‐	
   are	
   also	
   important	
   instruments,	
   in	
   their	
   own	
   right,	
   to	
  
accelerate	
  the	
  impact.	
  Our	
  scientists	
  are	
  increasingly	
  using	
  their	
  skills	
  and	
  their	
  
strategic	
   advantage	
   of	
   being	
  well	
   integrated	
   in	
   farming	
   communities	
   to	
   create	
  
well	
  designed	
  datasets	
  for	
  others	
  to	
  use.	
  While	
  there	
  is	
  no	
  doubt	
  that	
  progress	
  on	
  
development	
  related	
  research	
  questions	
  would	
  accelerate	
  with	
  increased	
  access	
  
to	
   well	
   documented	
   and	
   readily	
   accessible	
   data,	
   the	
   move	
   also	
   puts	
   more	
  
pressure	
  on	
  the	
  gatherers,	
  compilers,	
  analysts	
  and	
  keepers	
  of	
  the	
  data.	
  Because	
  
of	
  previous	
  emphasis	
  on	
  scientific	
  publications	
  only,	
  research	
  projects	
  usually	
  do	
  
not	
  budget	
  for	
  the	
  extra	
  costs	
  that	
  occur	
  for	
  publishing	
  and	
  long-­‐term	
  storage	
  of	
  
data.	
   While	
   projects	
   adequately	
   account	
   for	
   the	
   costs	
   of	
   conducting	
   research,	
  
including	
   the	
   collection	
   and	
   analysis	
   of	
   research	
   data,	
   they	
   seldom	
   include	
  
preparing	
  of	
  data	
  and	
  metadata	
  and	
  curation	
  as	
  part	
  of	
  the	
  costs	
  of	
  the	
  research	
  
process	
  led	
  alone	
  long-­‐term	
  costs	
  for	
  data	
  storage	
  and	
  preservation	
  beyond	
  the	
  
immediate	
   lifetime	
   of	
   a	
   project.	
   Even	
  more	
   importantly	
   the	
   research	
  nature	
   of	
  
CGIAR	
  makes	
  it	
  mandatory	
  to	
  ensure	
  that	
  the	
  contributions	
  of	
  those	
  individuals	
  
and	
  organizations	
  are	
  recognized,	
  whose	
  reputations	
  and	
  careers	
  depend	
  on	
  that	
  
recognition.	
   Data	
   generators	
   have	
   little	
   opportunities	
   to	
   gain	
   recognition	
   from	
  
PAGE	
  -­‐	
  3	
  -­‐	
  
publishing	
   high	
   value	
   datasets	
   under	
   a	
   performance	
   evaluation	
   system	
   that	
  
strongly	
  emphases	
  scientific	
  publications.	
  	
  
The	
   purpose	
   of	
   this	
   white	
   paper	
   is	
   to	
   provide	
   an	
   overview	
   of	
   the	
   ongoing	
  
initiatives	
  at	
  center	
   level	
  to	
  respond	
  to	
  changing	
  public	
  expectations	
  and	
  to	
  the	
  
challenge	
  of	
   improving	
   the	
   conduct	
   of	
   science	
  by	
  making	
   research	
  data	
  widely	
  
available.	
  We	
  also	
  attempt	
  to	
  provide	
  a	
  framework	
  for	
  implementing	
  open	
  access	
  
for	
  research	
  data	
  to	
  maximize	
  CGIAR’s	
  impact	
  on	
  development.	
  The	
  remainder	
  of	
  
this	
   paper	
   proceeds	
   as	
   follows;	
   firstly	
   a	
   summary	
   of	
   the	
   diversity	
   of	
   research	
  
data	
  produced	
  by	
   the	
   centers	
   is	
   given,	
   followed	
  by	
  an	
  overview	
  of	
   the	
  existing	
  
infrastructure	
   for	
   data	
   management	
   for	
   each	
   Center.	
   Secondly,	
   some	
   of	
   the	
  
limitations	
   and	
   barriers	
   faced	
   by	
   the	
   centers	
   in	
   their	
   process	
   to	
   mainstream	
  
research	
   data	
   publishing	
   are	
   addressed.	
   The	
   paper	
   concludes	
   with	
  
recommendations	
  for	
  how	
  these	
  limitations	
  and	
  barriers	
  can	
  be	
  tackled.	
  
Types	
  of	
  research	
  data	
  	
  
In	
   close	
   collaboration	
   with	
   national	
   research	
   institutions	
   almost	
   10,000	
  
scientists,	
   researchers,	
   technicians,	
   are	
   collecting,	
   analyzing	
   and	
   synthesizing	
  
data	
   on	
   smallholder	
   agricultural	
   systems	
   in	
   Asia,	
   Africa	
   and	
   Latin	
   America.	
  	
  
While	
   the	
   overall	
  mandate	
   of	
   the	
   CGIAR	
   Centers	
   is	
   the	
   same,	
   each	
   Center	
   has	
  
their	
   own	
   research	
   emphasis	
   resulting	
   in	
   a	
   vast	
   variety	
   of	
   different	
   kinds	
   of	
  
research	
   data.	
   	
   The	
   following	
   section	
   does	
   not	
   attempt	
   to	
   give	
   an	
   exhaustive	
  
description	
   of	
   the	
   data	
   we	
   produce,	
   but	
   rather	
   to	
   provide	
   an	
   overview	
   of	
   the	
  
main	
  types	
  and	
  their	
  characteristics.	
  
Long	
  term	
  trials	
  
Centers	
  that	
  are	
  focusing	
  on	
  breeding	
  of	
  food	
  crops	
  and	
  livestock	
  need	
  multiple	
  
location	
  trials	
  over	
  many	
  years	
  before	
  a	
  new	
  genotype	
  can	
  be	
  fully	
  evaluated,	
  so	
  
do	
   agroforestry	
   and	
   forestry	
   trials	
   that	
   are	
   working	
   with	
   slow	
   growing	
  
perennials.	
   Centers	
   working	
   more	
   generally	
   on	
   combined	
   systems	
   research,	
  
natural	
   resource	
  management	
   and	
   climate	
   change	
   also	
   have	
   research	
   covering	
  
decades	
  in	
  order	
  to	
  characterize	
  and	
  influence	
  changes	
  in	
  these	
  systems.	
  	
  	
  These	
  
kinds	
   of	
   data	
   are	
   often	
   collected	
   in	
   various	
   consecutive	
   projects	
   and	
   technical	
  
and	
  scientific	
  staff	
  that	
  collect	
  and	
  analyze	
  the	
  data	
  can	
  change.	
  From	
  a	
  scientific	
  
point	
   of	
   view	
   it	
   is	
   not	
   desirable	
   or	
   useful	
   to	
   make	
   the	
   data	
   available	
   before	
  
meaningful	
  intermediate	
  results	
  have	
  been	
  collected.	
  	
  
One-­‐off	
  data	
  collections	
  
Some	
  data	
  are	
  project	
  specific	
  and	
  aimed	
  at	
  answering	
  specific	
  research	
  and/or	
  
development	
   related	
   questions,	
   often	
   analyzed	
   and	
   published	
   as	
   part	
   of	
   the	
  
project	
   deliverables.	
   Publication	
   should	
   be	
   written	
   within	
   a	
   reasonable	
  
timeframe	
   after	
   which	
   the	
   data	
   can	
   be	
   publically	
   released	
   together	
   with	
   the	
  
publication.	
  What	
  is	
  a	
  reasonable	
  timeframe	
  does	
  not	
  only	
  vary	
  from	
  researcher	
  
to	
  researcher,	
  but	
  by	
  discipline.	
  Economics	
  publishing	
  moves	
  much	
  slower	
  than	
  
other	
  fields.	
  It	
  is	
  not	
  uncommon	
  for	
  a	
  paper	
  to	
  reach	
  publication	
  5-­‐10	
  years	
  after	
  
the	
  original	
  data	
  was	
  collected.	
  
PAGE	
  -­‐	
  4	
  -­‐	
  
Baseline	
  data	
  
Baseline	
   data,	
   either	
   household	
   surveys	
   or	
   biophysical	
   surveys	
   used	
   either	
   for	
  
basic	
   characterization	
   of	
   a	
   new	
  project	
   site,	
   or	
   for	
   an	
   impact	
   evaluation	
   of	
   the	
  
project.	
   The	
   data	
   are	
   collected	
   as	
   part	
   of	
   the	
   deliverables	
   of	
   the	
   project,	
   but	
  
budget	
  constraints	
  seldom	
  allow	
  sufficient	
  sample	
  size	
  or	
  rigid	
  sampling	
  designs	
  
to	
  allow	
  the	
  use	
  of	
  the	
  data	
  for	
  peer-­‐	
  reviewed	
  publications.	
  	
  These	
  data	
  usually	
  
get	
  analyzed	
  and	
  results	
  presented	
  in	
  donor	
  reports.	
  For	
  these	
  kinds	
  of	
  data	
  the	
  
curation	
  costs	
  are	
  actually	
  very	
  high	
  as	
  they	
  are	
  not	
  accompanied	
  by	
  a	
  scientific	
  
publication	
   that	
  provide	
   the	
  necessary	
  metadata	
  and	
  methodologies	
  are	
   stored	
  
in	
  various	
  versions	
  on	
  private	
  computers.	
  	
  The	
  data	
  has	
  limited	
  usability	
  for	
  the	
  
project	
  itself,	
  but	
  baseline	
  data	
  sets	
  combined	
  from	
  multiple	
  projects	
  potentially	
  
form	
  high	
  information	
  assets	
  for	
  the	
  institutes	
  and	
  the	
  public.	
  There	
  is	
  a	
  tradeoff	
  
between	
  releasing	
  the	
  data	
  to	
  do	
  comparative	
  analysis	
  or	
  to	
  wait	
  until	
  the	
  follow	
  
up	
  study	
  is	
  completed.	
  
Genomic	
  data	
  
Some	
   Centers	
   are	
   involved	
   in	
   genome	
   sequencing	
   projects,	
   specifically	
   in	
  
generating	
   the	
   NGS	
   (next	
   generation	
   sequencing)	
   data	
   for	
   gene-­‐phenotype	
  
association	
   studies.	
   These	
   studies	
   can	
   generate	
   multi-­‐terabases	
   of	
   sequencing	
  
data.	
  One	
  of	
  the	
  key	
  challenges	
  is	
  to	
  devise	
  scalable	
  and	
  robust	
  data	
  management	
  
and	
   data	
   sharing	
   solutions.	
   	
   High-­‐performance	
   computing	
   and	
   storage	
   are	
  
required	
  to	
  efficiently	
  process	
  data	
  generated	
  by	
  NGS.	
  Bioinformatics	
  support	
  is	
  
integral	
   to	
   address	
   data	
   management	
   systems	
   dealing	
   with	
   efficient	
   storage,	
  
retrieval,	
   data	
  mining,	
  data	
  analysis	
   and	
  making	
  data	
  available	
   to	
   the	
  public	
   at	
  
the	
  appropriate	
  time.	
  
Data	
  collected	
  as	
  part	
  of	
  a	
  research	
  thesis	
  
Various	
   projects	
   have	
   a	
   specific	
   capacity	
   building	
   requirement,	
   whereby	
  
postgraduate	
   students	
   collect	
   a	
   substantial	
   part	
   of	
   the	
   data.	
   	
   The	
   data	
   is	
   to	
   be	
  
used	
   in	
   peer-­‐reviewed	
   publications	
   as	
   partial	
   requirement	
   for	
   their	
   degrees.	
  	
  
Data	
  can	
  only	
  be	
  publically	
  released	
  after	
  the	
  student	
  has	
  published	
  their	
  papers	
  
or	
  their	
  thesis.	
  	
  
Value-­‐	
  added	
  secondary	
  datasets	
  
A	
  large	
  proportion	
  of	
  the	
  work	
  of	
  a	
  CGIAR	
  scientist	
  is	
  to	
  review	
  and	
  analyze	
  pre-­‐
existing	
   data	
   that	
  was	
   not	
   gathered	
   or	
   collected	
   by	
   the	
   authors	
   of	
   the	
   current	
  
research	
  project.	
  Usually	
  it	
  has	
  been	
  collected	
  by	
  another	
  organization	
  or	
  source	
  
or	
  data	
  collected	
  from	
  government	
  publications.	
  Typical	
  secondary	
  datasets	
  that	
  
are	
  used	
  are	
  meteorological	
  datasets,	
  remotely	
  sensed	
  data	
  often	
  in	
  the	
  form	
  of	
  
satellite	
  images	
  or	
  aerial	
  photographs,	
  panel	
  data	
  sets	
  for	
  rural	
  households.	
  	
  The	
  
secondary	
   datasets	
   are	
   shared	
   with	
   CGIAR	
   scientists	
   under	
   specific	
   user	
  
agreements	
   or	
   licenses	
   and	
   cannot	
   be	
   publically	
   shared	
   by	
   the	
   scientists.	
   Data	
  
products	
  derived	
  from	
  these	
  data	
  can	
  be	
  shared	
  however,	
  without	
  the	
  raw	
  data.	
  
Spatial	
  data	
  
While	
   most	
   data	
   generated	
   in	
   the	
   CGIAR	
   research	
   activities	
   has	
   a	
   spatial	
  
component	
  (with	
  exception	
  of	
  pure	
  lab	
  analysis	
  not	
  related	
  to	
  specific	
  locations)	
  
the	
   GIS	
   units	
   of	
   the	
   individual	
   centers	
   collect,	
   transform	
   and	
   generate	
   a	
   large	
  
amount	
  of	
  spatial	
  data.	
  These	
  fall	
  into	
  three	
  categories:	
  
PAGE	
  -­‐	
  5	
  -­‐	
  
a. Spatial	
  data	
   (in	
   form	
  of	
  polygons	
  and	
   raster,	
   satellite	
   imagery)	
  obtained	
  
from	
   other	
   parties	
   like	
   NASA,	
   JRC,	
   National	
   Geographical	
   Institutes,	
  
Universities	
  and	
  other	
  research	
  institutions,	
  NGOs,	
  private	
  companies	
  etc.	
  
This	
  is	
  used	
  for	
  analysis,	
  mapping,	
  targeting	
  within	
  the	
  work	
  of	
  the	
  units.	
  
Often	
   this	
   original	
   data	
   is	
   curated	
   and	
   improved	
   and	
   represents	
   a	
   new	
  
international	
   public	
   good	
   and	
   falls	
   under	
   value-­‐added	
   secondary	
  
datasets.	
  	
  
b. Geo	
   referenced	
  data	
   generated	
   by	
   projects	
   and	
   the	
  GIS	
   units	
  within	
   the	
  
centers.	
   This	
   can	
   be	
   any	
   other	
   form	
   of	
   data	
   collected	
   such	
   as	
   socio	
  
economic	
  surveys,	
  soil	
  samples,	
  germplasm	
  collections	
  etc.	
  It	
  is	
  then	
  often	
  
combined	
  with	
  other	
  spatial	
  data	
  sets	
  as	
  mentioned	
  under	
  a)	
  and	
  utilized	
  	
  
for	
  further	
  analysis.	
  
c. Statistical	
  	
  and	
  other	
  related	
  data	
  (climate	
  from	
  stations)	
  that	
  is	
  collected	
  
from	
  national	
  and	
  other	
  institutions	
  such	
  as	
  subnational	
  crop	
  production	
  
data	
  or	
  poverty	
  statistics.	
  This	
  is	
  georeferenced,	
  often	
  extrapolated	
  and	
  if	
  
necessary	
   further	
   disaggregated	
   and	
   converted	
   into	
   common	
   GIS	
   data	
  
formats	
   and	
   made	
   available	
   to	
   the	
   public	
   and	
   other	
   centers	
   again	
   as	
  
value-­‐added	
  secondary	
  datasets.	
  	
  
Data	
  collected	
  in	
  a	
  private	
  public	
  partnership	
  project	
  
When	
   working	
   with	
   private	
   companies	
   some	
   of	
   the	
   data	
   and	
   information	
   is	
  
highly	
   sensitive	
   and	
   usually	
   the	
   centers	
   sign	
   confidentiality	
   agreements	
   that	
  
state	
   explicitly	
  what	
   the	
   data	
   can	
   be	
   used	
   for	
   and	
  which	
   data	
   products	
   can	
   be	
  
made	
  publically	
  available.	
  
R	
  &	
  D	
  Datasets	
  
CGIAR	
   has	
   programs	
   that	
   are	
   designed	
   and	
   funded	
   purposely	
   to	
   create	
   public	
  
databases	
   on	
   key	
   agricultural	
   indicators.	
   	
   The	
   databases	
   are	
   dynamic	
   and	
   are	
  
updated	
   on	
   regular	
   intervals.	
   Technical	
   and	
   scientific	
   staff	
   that	
   collect	
   and	
  
analyze	
   the	
   data	
   can	
   change.	
   Here	
   only	
   aggregated	
   data	
   are	
   made	
   available	
  
together	
  with	
  the	
  methods.	
  
Partner	
  data	
  
Most	
  of	
  our	
  projects	
  are	
  done	
  together	
  with	
  partners	
  that	
  sometimes	
  contribute	
  
their	
  own	
  non-­‐CGIAR	
  funded	
  data.	
  Here	
  it	
  is	
  important	
  that	
  the	
  partner	
  decides	
  
what	
  the	
  data	
  should	
  be	
  used	
  for	
  and	
  who	
  should	
  have	
  access	
  to	
  the	
  data.	
  Often	
  
the	
  partner	
  has	
  only	
  agreed	
  to	
  joint	
  copyright	
  of	
  data	
  products,	
  but	
  not	
  the	
  actual	
  
data	
  sets.	
  	
  
Research	
  data	
  infrastructure	
  across	
  the	
  centers	
  
While	
   publishing	
   research	
   data	
   as	
   a	
   research	
   output	
   is	
   nothing	
   new	
   for	
   the	
  
CGIAR	
   Centers,	
   in	
   the	
   past	
   it	
   was	
   confined	
   to	
   flagship	
   programs	
   and	
   projects.	
  	
  
Data	
   sharing	
   agreements	
   and	
   data	
   management	
   policies	
   were	
   developed	
   and	
  
data	
   managers	
   and	
   curators	
   hired	
   based	
   on	
   the	
   individual	
   needs	
   of	
   these	
  
projects.	
  Centers	
  did	
  not	
  have	
  the	
  infrastructure,	
  capacity	
  and	
  incentive	
  to	
  make	
  
more	
  of	
   their	
   research	
  data	
  widely	
  available.	
   In	
  early	
  2008	
  a	
   first	
   attempt	
  was	
  
made	
  by	
  the	
  Alliance	
  Deputy	
  Executive	
  (ADE)	
  to	
  review	
  data	
  management	
  across	
  
the	
   centers	
   and	
  explore	
  mechanisms	
   for	
   strengthening	
   collective	
  action	
  among	
  
PAGE	
  -­‐	
  6	
  -­‐	
  
centers	
   (Anon,	
   2008).	
   A	
   lot	
   has	
   changed	
   since	
   then	
   and	
   Center	
  wide	
   research	
  
data	
   management	
   policies	
   are	
   becoming	
   the	
   norm	
   rather	
   than	
   the	
   exception.	
  
Increasing	
   investment	
   in	
   research	
   support	
   units	
   and	
   or	
   staff	
   and	
   the	
  
implementation	
   of	
   data	
   archives	
   reflect	
   the	
   change	
   of	
   mind	
   set	
   within	
   center	
  
management	
  as	
  well	
  scientists.	
  	
  Most	
  centers	
  have	
  already	
  or	
  are	
  in	
  the	
  process	
  
of	
  setting-­‐up	
  'OAI-­‐compliant'1	
  data	
  repositories,	
  which	
  allow	
  the	
  easy	
  harvest	
  of	
  
metadata	
  from	
  one	
  repository	
  to	
  another.	
   	
  Table	
  1	
  provides	
  a	
  brief	
  summary	
  of	
  
the	
  current	
   research	
  data	
   infrastructure	
  across	
   centers.	
  Details	
   for	
  each	
  Center	
  
are	
  given	
  below.	
  
	
  
Table	
   1:	
   Overview	
   of	
   existing	
   infrastructure	
   for	
   research	
   data	
   management	
   &	
  
bioinformatics	
  across	
  the	
  different	
  institutes.	
  
Research	
   Centralized	
  
data	
   Data	
   Data	
  
Management	
   Management	
   Geoinformatics	
   Biometrics	
   Archiving	
  &	
  
Centre	
   Policy	
   Unit	
   Unit	
   Unit	
   sharing	
  
Africa	
  Rice	
   YES	
   YES	
   YES	
   YES	
   Since	
  July	
  2012	
  
Bioversity	
   In	
  process	
   YES	
   	
   	
   Since	
  Sept.	
  2013	
  
CIAT	
  	
   YES	
   YES	
   YES	
   	
   In	
  process	
  
CIFOR	
   YES	
   	
   	
   	
   In	
  process	
  
CIMMYT	
   YES	
   Recruiting	
   YES	
   YES	
   In	
  process	
  
CIP	
   YES	
   YES	
   YES	
   YES	
   YES	
  
ICARDA	
   In	
  process	
   YES	
   YES	
   YES	
   In	
  process	
  
ICRAF	
   YES	
   YES	
   YES	
   YES	
   Since	
  2011	
  
ICRISAT	
   In	
  process	
   YES	
   YES	
   YES	
   YES	
  
IFPRI	
   YES	
   Recruiting	
   No	
   	
   Since	
  2005	
  
IITA	
   In	
  process	
   In	
  process	
   YES	
   YES	
   In	
  process	
  
Partial	
  
(shared	
  
ILRI	
   YES	
   YES	
   YES	
   YES	
   servers,	
  data	
  
portal	
  in	
  
development)	
  
YES	
  
IRRI	
   (Currently	
  being	
   YES	
   YES	
   YES	
   In	
  process	
  	
  
updated)	
  
IWMI	
   YES	
   	
   YES	
   	
   YES	
  
World	
  Fish	
   YES	
   YES	
   YES	
   	
   YES	
  
	
  
	
   	
  
	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
1	
  The	
  Open	
  Archives	
  Initiative	
  (OAI)	
  develops	
  and	
  promotes	
  interoperability	
  standards	
  (Protocol	
  
for	
  Metadata	
  Harvesting)	
  that	
  aim	
  to	
  facilitate	
  the	
  efficient	
  dissemination	
  of	
  content.	
  	
  
PAGE	
  -­‐	
  7	
  -­‐	
  
Africa	
  Rice	
  
AfricaRice	
  research	
  support	
  unit,	
  the	
  Data	
  Integration	
  and	
  Biometrics	
  Unit	
  (DIB	
  
Unit)	
  is	
  responsible	
  for	
  data	
  management	
  and	
  to	
  assist	
  AfricaRice	
  research	
  staff	
  
(and	
   partners)	
   to	
   enhance	
   the	
   efficiency	
   and	
   efficacy	
   of	
   data	
   management	
  
processes:	
   i)	
   data	
   acquisition,	
   quality	
   control	
   and	
   storage:	
   backstopping	
   of	
  
AfricaRice	
  research	
  staff	
  in	
  the	
  management	
  of	
  their	
  experimental	
  data,	
  including	
  
automating	
  data	
   gathering	
  procedures	
   and	
  quality	
   control;	
   ii)	
   data	
   integration,	
  
analysis	
  and	
  visualization:	
  biometrics’	
  advice	
  and	
  support	
  to	
  AfricaRice	
  research	
  
staff	
   on	
   experimental	
   designs,	
   GxE	
   analyses,	
   mapping	
   quantitative	
   trait	
   loci	
  
(QTLs),	
  analyzing	
  genetic	
  diversity,	
  development	
  of	
  decision	
  support	
  tools,	
  etc.	
  In	
  
addition,	
   the	
   DIB	
   unit	
   provides	
   institute-­‐wide	
   data	
   standards,	
   including	
   a	
  
common	
   vocabulary	
   and	
   the	
   use	
   of	
   standardized	
   formats	
   for	
   primary	
   and	
  
metadata.	
   It	
   enables	
   AfricaRice	
   and	
   partners	
   to	
   more	
   easily	
   integrate,	
  
synchronize	
  and	
  consolidate	
  data	
   from	
  different	
  programs,	
   exchange	
  data	
  with	
  
other	
  organizations	
   in	
  a	
   common	
   format,	
   and	
  communicate	
  effectively	
   through	
  
shared	
  terms	
  and	
  reporting	
  formats.	
  
	
  The	
  Unit	
  consists	
  of	
  two	
  statisticians	
  (one	
  IRS	
  and	
  one	
  Support	
  Staff),	
  one	
  data	
  
manager	
  (Support	
  Staff)	
  and	
  one	
  Consultant	
  (Support	
  Staff).	
  Africa	
  Rice	
  is	
  using	
  
Dataverse	
  as	
  its	
  research	
  data	
  archive.	
  It	
  was	
  released	
  on	
  July	
  2012	
  and	
  consists	
  
of	
  23	
  studies	
  as	
  of	
  today.	
  AfricaRice	
  developed	
  a	
  Data	
  Management	
  and	
  Sharing	
  
Policy	
  in	
  1999.	
  Revised	
  in	
  2013,	
  the	
  document	
  is	
  in	
  the	
  process	
  for	
  approbation	
  
by	
  the	
  AfricaRice	
  Board	
  of	
  Trustees.	
  
Bioversity	
  
Bioversity	
   has	
   been	
   engaged	
   in	
   developing	
   the	
   CGIAR	
  Open	
   Access	
   policy	
   and	
  
guidelines	
  together	
  with	
  other	
  CGIAR	
  Centers.	
  Once	
  approved,	
  these	
  documents	
  
will	
  form	
  the	
  basis	
  for	
  Bioversity’s	
  ‘Open	
  Access	
  and	
  Data	
  Management	
  Plan’.	
  We	
  
have	
  already	
  drafted	
  this	
  plan	
  based	
  on	
  the	
  current	
  version	
  of	
  the	
  OA	
  policy,	
  and	
  
we	
  hope	
  to	
  have	
  the	
  policy	
  approved	
  in	
  early	
  2014.	
  	
  
The	
  responsibility	
  for	
  the	
  collection	
  and	
  management	
  of	
  research	
  data	
  is	
  with	
  the	
  
Research	
  Planning	
   and	
  Monitoring	
  Unit	
   at	
  Bioversity.	
   	
   The	
  Library	
   staff	
  within	
  
this	
  unit	
  are	
  responsible	
  for	
  1)	
  collecting	
  the	
  datasets;	
  2)	
  ensuring	
  completeness	
  
and	
   accuracy	
   of	
   the	
   associated	
   metadata;	
   3)	
   finding/linking	
   the	
   datasets	
   with	
  
associated	
  publications	
  and	
  tools,	
  and	
  4)	
  inputting/releasing	
  it	
  on	
  the	
  Dataverse	
  
network.	
   	
  No	
  extra	
  personnel	
  have	
  been	
  recruited	
  for	
  these	
  tasks	
  so	
  far.	
  Whilst	
  
Bioversity	
  does	
  not	
  have	
  a	
  GIS	
  unit,	
  there	
  are	
  3	
  GIS	
  specialists	
  based	
  in	
  regional	
  
offices	
  and	
  HQ.	
  	
  
Bioversity	
   recently	
   established	
   an	
   open	
   access	
   dataset	
   repository	
   on	
   the	
  
Dataverse	
  (Harvard)	
  Platform.	
  	
  	
  At	
  present	
  we	
  have	
  datasets	
  available	
  there	
  from	
  
research	
  carried	
  out	
  in	
  2012,	
  and	
  we	
  are	
  currently	
  collecting	
  other	
  datasets	
  from	
  
previous	
  years.	
  At	
  present,	
  the	
  metadata	
  of	
  each	
  dataset	
  is	
  available	
  immediately	
  
but	
  the	
  actual	
  datasets	
  are	
  only	
  released	
  once	
  the	
  scientist	
  gives	
  their	
  approval.	
  	
  
In	
  order	
  to	
  collect	
  datasets	
  systematically	
  Bioversity	
  has	
  in	
  place	
  focal	
  points	
  in	
  
its	
  5	
  research	
  programs	
  that	
  assist	
  the	
  Library	
  staff	
  primarily	
  with	
  collecting	
  the	
  
datasets	
  from	
  their	
  scientific	
  staff.	
  In	
  collaboration	
  with	
  partners	
  Bioversity	
  also	
  
manages	
  a	
  number	
  of	
  specialized	
  databases	
  and	
  datasets	
  relating	
  to	
  agricultural	
  
and	
  forest	
  biodiversity	
  such	
  as:	
  
	
  
	
  
PAGE	
  -­‐	
  8	
  -­‐	
  
• Collecting	
  Mission	
  database	
  (http://bioversity.github.io/geosite/)	
  	
  
• Musa	
  Germplasm	
  Information	
  System	
  (MGIS)	
  	
  (http://www.crop-­‐
diversity.org/banana/)	
  
• EURISCO	
  	
  (http://eurisco.ecpgr.org/nc/home_page.html)	
  
• New	
  World	
  Fruit	
  Database	
  (http://nwfdb.bioversityinternational.org/)	
  
Bioversity,	
   always	
   in	
   collaboration	
   with	
   its	
   partners,	
   also	
   develops	
   data	
  
standards	
  for	
  documentation	
  and	
  protocols	
  to	
  enable	
  information	
  sharing.	
  	
  
The	
   Descriptor	
   Lists	
   publications	
   that	
   are	
   published	
   by	
   Bioversity,	
   assist	
  
researchers	
  and	
  genebank	
  curators	
  to	
  improve	
  their	
  capacity	
  to	
  describe,	
  store,	
  
manage	
   and	
   share	
   information	
   about	
   plant	
   resources,	
   whether	
   stored	
  
in	
  genebanks	
  or	
   growing	
   in	
   their	
   natural	
   environments	
   (e.g.	
   FAO/Bioversity	
  
International,	
  2012).	
  
Center	
  for	
  International	
  Forestry	
  Research	
  (CIFOR)	
  
CIFOR’s	
   Director	
   General	
   approved	
   and	
   promulgated	
   a	
   Research	
   Data	
  
Management	
  Policy	
   and	
  Guidelines	
   and	
  Procedures	
   in	
   July	
   2013.	
   These	
   identify	
  
project	
   managers	
   as	
   having	
   primary	
   responsibility	
   for	
   management,	
   including	
  
archiving,	
  of	
  research	
  data;	
  require	
  preparation	
  and	
  implementation	
  of	
  Research	
  
Data	
  Management	
  Plans;	
  and	
  identify	
  the	
  Center’s	
  Data	
  and	
  Information	
  Services	
  
Unit	
  as	
  having	
  responsibility	
  for	
  ensuring	
  storage	
  of	
  and	
  access	
  to	
  archived	
  data,	
  
with	
  support	
   from	
  the	
  GIS	
  Unit	
   for	
  spatial	
  data.	
  CIFOR	
   is	
  currently	
  recruiting	
  a	
  
Data	
  Librarian	
   to	
  play	
  a	
  primary	
   role	
   in	
   supporting	
  project	
  managers	
  and	
  staff	
  
implementing	
  the	
  Policy.	
  	
  
International	
  Potato	
  Center	
  (CIP)	
  
Around	
  September	
  2002,	
  CIPs	
  director	
  of	
   research	
   implemented	
  a	
   service	
  unit	
  
charged	
   with	
   building	
   and	
   maintaining	
   the	
   institutional	
   'memory'	
   of	
   research	
  
data	
  (todays	
  rough	
  equivalent	
  of	
  KM	
  on	
  scientific	
  data).	
  To	
  this	
  end	
  it	
  provides	
  
both	
   centralized	
   archiving,	
   documentation	
   and	
   databasing	
   services	
   as	
   well	
   as	
  
support	
  to	
  researchers	
  in	
  data	
  analysis	
  and	
  visualization.	
  This	
  includes	
  also	
  the	
  
de-­‐novo	
   production	
   of	
   databases	
   and	
   tools	
   as	
   appropriate	
   versus	
   re-­‐using	
  
existing	
   software.	
   The	
   latter	
   aspect	
   was	
   seen	
   as	
   intrinsic	
   to	
   the	
   concept	
   of	
   a	
  
'memory'	
   since	
   no	
  memory	
   'makes	
   sense'	
  without	
   'data'	
   and	
   'data	
   processing'	
  
tools.	
  Since	
  the	
  beginning,	
  the	
  unit	
  has	
  been	
  involved	
  in	
  developing	
  community	
  
data	
   documentation	
   standards	
   as	
   well	
   as	
   in	
   the	
   application	
   of	
   open-­‐source	
  
software	
  principles	
  and	
  the	
  promotion	
  of	
  open	
  access.	
  The	
  unit	
  was	
  called	
  at	
  the	
  
time	
  'research	
  informatics	
  unit	
  (RIU)'	
  and	
  recently	
  renamed	
  to	
  'Integrated	
  IT	
  and	
  
Computational	
  Research.	
  The	
  unit	
  provides	
  services	
  in	
  documentation,	
  database	
  
design	
  and	
  development,	
  PC-­‐based	
  and	
  mobile	
  application	
  development,	
  GIS	
  and	
  
molecular	
   bioinformatics.	
   The	
   unit	
   has	
   been	
   part	
   of	
   former	
   CGIAR	
   KM-­‐related	
  
activities	
  like:	
  the	
  IPD	
  (the	
  Intergenebank	
  Potato	
  Database,	
  a	
  database	
  to	
  cross-­‐
reference	
   potato	
   collections	
   and	
   share	
   passport	
   and	
   evaluation	
   data,	
   since	
  
~1995),	
  the	
  SINGER	
  community	
  as	
  a	
  repository	
  of	
  long-­‐term	
  data	
  on	
  germplasm	
  
(since	
  end	
  of	
  1990ies;	
  now	
  moving	
  towards	
  Genesys-­‐2	
  software	
  from	
  the	
  Global	
  
Trust),	
   the	
   GenerationCP	
   on	
   breeding	
   materials	
   including	
   the	
   'composite	
  
genotype	
  sets',	
  and	
  the	
  ICT-­‐KM	
  initiative.	
  
PAGE	
  -­‐	
  9	
  -­‐	
  
International	
  Center	
  for	
  Agricultural	
  Research	
  in	
  the	
  Dry	
  Areas	
  (ICARDA)	
  
ICRADA	
   has	
   a	
   Geoinformatics	
   Unit	
   for	
   all	
   the	
   geospatial	
   database	
   and	
   related	
  
products	
   in	
   coordination	
   with	
   Biometrics,	
   GRS,	
   CODIS	
   etc.	
   At	
   present	
   data	
   is	
  
stored	
  in	
  several	
  individual	
  archives	
  and	
  the	
  institute	
  is	
  in	
  the	
  process	
  of	
  setting	
  
up	
   centralized	
  data	
   archiving	
   facility	
   for	
   collecting,	
   streamlining,	
   archiving	
   and	
  
sharing,	
   supported	
   by	
   a	
   research	
   data	
   management	
   policy.	
   Less	
   than	
   10%	
   of	
  
research	
  data	
  is	
  currently	
  accessible	
  through	
  the	
  public	
  domain.	
  
International	
  Center	
  for	
  Tropical	
  Agriculture	
  (CIAT)	
  
CIAT	
  strongly	
  believes	
  that	
  open	
  access	
  to	
  data	
  produced	
  by	
  Staff	
  and	
  partners	
  
strategically	
   supports	
   our	
   mission	
   and	
   ensures	
   transparency	
   and	
   equity	
   in	
  
exploitation	
   of	
   the	
   opportunities	
   created.	
  	
   Even	
   before	
   implementing	
   a	
   Data	
  
Management	
   policy	
   in	
   February	
   2012,	
   CIAT	
   led	
   sharing	
   of	
   geo-­‐spatial	
   data	
  
together	
  with	
  associated	
  information,	
  at	
  different	
  scale	
  in	
  an	
  organized,	
  standard	
  
and	
   consistent	
  way.	
  With	
   the	
   implementation	
   of	
   the	
   Data	
  management	
   policy,	
  
CIAT	
  is	
  in	
  the	
  process	
  of	
  putting	
  40	
  years	
  of	
  data	
  from	
  the	
  institutional	
  memory	
  
(includes	
  phenotypic,	
  genotypic,	
  climatic,	
  spatial	
  and	
  socio-­‐economic	
  data)	
  in	
  the	
  
public	
   domain.	
  	
   The	
   work	
   involves	
   compilation	
   of	
   data	
   from	
   the	
   institutional	
  
memory,	
   cleaning	
   and	
   re-­‐formatting,	
   respecting	
   current	
   data	
   ontologies	
   and	
  
standards,	
  and	
  publication	
  through	
  appropriate	
  mechanisms.	
  	
  
While	
   CIAT	
   encourage	
   the	
   data	
   sharing	
   culture,	
   data	
   management	
   function	
   is	
  
currently	
   spread	
   across	
   research	
   areas	
   and	
   different	
   units.	
   The	
   information	
  
technology	
   unit	
   is	
   in	
   charge	
   of	
   data	
   storage.	
   Starting	
   2014,	
   we	
   expect	
   to	
  
consolidate	
   some	
   of	
   the	
   ongoing	
   activities	
   under	
   Library	
   and	
   information	
  
management.	
  In	
  addition,	
  CIAT	
  also	
  hosts	
  the	
  CCAFS	
  data	
  manager.	
  	
  
Most	
  of	
  the	
  field	
  data	
  are	
  now	
  collected	
  using	
  a	
  hand	
  held	
  PDA,	
  which	
  facilitates	
  
easy	
   backup	
   and	
   consolidation.	
  We	
   are	
   beginning	
   to	
   use	
   a	
   CIAT	
   Dataverse,	
   to	
  
facilitate	
  data	
  sharing	
  associated	
  with	
  research	
  publications.	
  Effort	
  is	
  ongoing	
  to	
  
share	
  trial	
  data	
  from	
  commodity	
  programs	
  via	
  Agtrials.	
  Given	
  the	
  very	
  large	
  size	
  
of	
   bioinformatics	
   and	
   genomic	
   data	
   sets,	
  further	
   discussion	
   is	
   ongoing	
   to	
   see	
  
how/	
  and	
  if	
  we	
  share	
  terra	
  bites	
  of	
  raw	
  data	
  and	
  mechanisms	
  for	
  doing	
  that	
   in	
  
terms	
  of	
  server	
  capacity	
  etc.	
  	
  Quite	
  a	
  lot	
  of	
  data	
  is	
  already	
  made	
  available	
  through	
  
GCP	
  https://www.integratedbreeding.net/.	
  	
  
International	
  Crops	
  Research	
  Institute	
  for	
  the	
  Semi-­‐Arid	
  Tropics	
  (ICRISAT)	
  
Recognizing	
   the	
   value	
   of	
   research	
   data	
   aggregation,	
   analysis	
   and	
   availability,	
  
ICRISAT	
   established	
   a	
   Data	
   Management	
   Unit	
   in	
   August	
   2010	
   to	
   mainstream,	
  
support	
   and	
  manage	
   research	
   data	
   for	
   its	
   preservation	
   and	
   publication	
   across	
  
the	
  institute	
  and	
  to	
  the	
  extent	
  possible,	
  to	
  make	
  the	
  data	
  publicly	
  available	
  for	
  all	
  
users	
  to	
  use.	
  The	
  unit	
  consists	
  of	
  a	
  full	
  time	
  Senior	
  Data	
  Manager	
  supported	
  by	
  
consultants,	
   research	
   fellows	
  and	
  Biometric	
  unit	
   of	
   ICRISAT.	
   	
  Biometric	
  unit	
   at	
  
ICRISAT	
  work	
  on	
  data	
  quality	
  activities	
  with	
  a	
  full	
  time	
  biometrician	
  along	
  with	
  
five	
  statisticians	
  and	
  research	
  scholars.	
  	
  
The	
  DMU	
  of	
  ICRISAT	
  have	
  come	
  up	
  with	
  various	
  data	
  management	
  platforms	
  for	
  
better	
   pedigree	
   management,	
   breeding	
   practice	
   analysis,	
   survey	
   management,	
  
climate	
  prediction	
  activities,	
  etc.	
  ICRISAT	
  has	
  also	
  introduced	
  several	
  innovative	
  
platforms	
   to	
   its	
   scientists	
   that	
   includes	
   open	
   as	
   well	
   as	
   commercial	
   such	
   as	
  
Agrobase,	
  aWhere	
  (cloud	
  based	
  data	
  management	
  tool),	
  VDSA	
  (data	
  warehouse),	
  
IBP	
  etc.	
  and	
  also	
  developed	
  several	
  applications	
  that	
  helps	
  better	
  visualization	
  &	
  
PAGE	
  -­‐	
  10	
  -­‐	
  
sharing	
  of	
  research	
  data	
  with	
  GIS	
  analysis	
  capabilities.	
  The	
  DMU	
  also	
  defined	
  and	
  
developed	
   workflows	
   and	
   protocols	
   for	
   managing	
   research	
   data	
   that	
   being	
  
produced	
  by	
  various	
  research	
  programs.	
  Training	
  programs	
  and	
  workshops	
  have	
  
been	
   organized	
   to	
   the	
   scientists	
   and	
   partners	
   working	
   on	
   ICRISAT’s	
   research	
  
programs	
  on	
  use	
  of	
  new	
  tools	
  and	
  data	
  management	
  platforms.	
  ICRISAT is	
  using	
  
Dataverse	
  as	
  a	
  platform to publish datasets. Nearly about 400 datasets in 5 formats in 
42 study areas have been uploaded in to ICRISAT-Dataverse application. ICRISAT	
  
has	
  adopted	
  Open	
  Access	
  Policy	
  and	
  launched	
  repository	
  in	
  May	
  2011	
  to	
  provide	
  
an	
  easy	
  interface	
  for	
  researchers,	
  practitioners,	
  or	
  web-­‐connected	
  farmers	
  to	
  use,	
  
build	
  on	
  and	
  share	
  research	
  conducted	
  at	
  ICRISAT. 
DMU is working with a strategy to manage the research data	
   that	
  being	
  produced	
  
across	
  the	
  institute,	
  covering	
  all	
  the	
  locations	
  and	
  thematic/program	
  areas;	
  and	
  
organize	
  it	
  efficiently	
  in	
  a	
  “central	
  data	
  repository”	
  for	
  allowing	
  Global	
  scientific	
  
community	
  “access	
  to	
  the	
  data	
  to	
  get	
  successful	
  results”	
  for	
  addressing	
  pressing	
  
Global	
  issues.	
  	
  	
  Efforts	
  have	
  also	
  been	
  made	
  to	
  bring	
  a	
  cultural	
  change	
  at	
  scientist	
  
level	
  by	
  continuous	
  interactions	
  and	
  by	
  educating	
  them	
  on	
  the	
  advantages	
  of	
  the	
  
data	
  sharing.	
  This	
  is	
  a	
  win-­‐win	
  situation	
  for	
  the	
  institutes	
  as	
  well	
  as	
  the	
  scientific	
  
community.	
  For	
  this,	
  ICRISAT	
  has	
  come	
  up	
  with	
  Data	
  Management	
  Policy	
  and	
  has	
  
been	
  put	
  forward	
  for	
  the	
  research	
  committee	
  approvals.	
  
International	
  Food	
  Policy	
  Research	
  Institute	
  (IFPRI)	
  
IFPRI’s	
  Communication	
  and	
  Knowledge	
  Management	
  Division,	
  has	
  a	
  Knowledge	
  
Management	
  Unit	
  considered	
  a	
  support	
  unit	
  for	
  research.	
  	
  The	
  KM	
  unit	
  provides	
  
support	
   for	
   preparing	
   the	
   data	
   documentation,	
   data	
   curation,	
   creates	
   the	
  
connection	
   with	
   associated	
   publications	
   and	
   tools,	
   monitors	
   it	
   uses	
   and	
   also	
  
serves	
  as	
  a	
  user	
  support	
  and	
  Q&A	
  center.	
  	
  	
  At	
  present	
  the	
  unit	
  has	
  1	
  full	
  time	
  data	
  
curator	
  (with	
  research	
  background),	
  a	
  half	
  time	
  knowledge	
  manager,	
  that	
  helps	
  
with	
  the	
  taxonomies	
  and	
  metadata	
  and	
  citation	
  statistics.	
  
The	
  Communications	
  and	
  Knowledge	
  Management	
  Division	
  is	
  pursuing	
  to	
  hire	
  a	
  
Data	
  Manager,	
  who	
  will	
  provide	
  support	
  during	
  the	
  research	
  cycle,	
  development	
  
of	
   data	
   plans	
   when	
   the	
   project	
   is	
   conceived,	
   conceptual	
   frameworks,	
   models,	
  
questionnaires	
  and	
  analysis.	
  	
  	
  	
  
IFPRI	
  has	
  a	
  Dataset	
  Policy	
  existing	
  since	
  2000,	
  was	
  updated	
   in	
  2010,	
  making	
   it	
  
mandatory	
   for	
   researchers	
   to	
   provide	
   open	
   access	
   to	
   the	
   relevant	
   data,	
   while	
  
safeguarding	
   the	
   privacy	
   of	
   participants	
   and	
   protecting	
   confidential	
   and	
  
proprietary	
  information.	
  	
  IFPRI	
  will	
  make	
  all	
  primary	
  and	
  value-­‐added	
  secondary	
  
datasets	
  collected	
  after	
  January	
  1,	
  1999,	
  publicly	
  available	
  two	
  (2)	
  years	
  after	
  all	
  
data	
  collection	
  ceases	
  or,	
  before	
  two	
  years,	
  at	
  the	
  time	
  of	
  a	
  major	
  publication	
  by	
  
the	
  lead	
  data	
  collector.	
   	
  The	
  datasets	
  are	
  released	
  after	
  the	
  lead	
  researcher	
  and	
  
Division	
  Director	
  has	
  approved	
  it.	
  
In	
  2012,	
  11	
  datasets	
  were	
  prepared	
  for	
  posting	
  in	
  the	
  open	
  repository	
  at	
  IFPRI	
  
Dataverse,	
   and	
  now	
  we	
  have	
   a	
   total	
   of	
   97	
  datasets.	
   	
   These	
   datasets	
   have	
   been	
  
cited	
  120	
  times	
  (according	
  to	
  ISI	
  data	
  citation	
  index)	
  and	
  have	
  been	
  downloaded	
  
19,289	
  times.	
  Whenever	
  possible	
  we	
  linked	
  the	
  publication	
  to	
  dataset	
  and	
  vice-­‐
versa	
   so	
   that	
   the	
   discoverability	
   of	
   dataset	
   and	
   related	
   publications	
   becomes	
  
easier.	
   Preparing	
   a	
   dataset	
   for	
   the	
   open	
   repository,	
   takes	
   between	
   2.5-­‐6	
   days	
  
depending	
  on	
  how	
  the	
  raw	
  data	
  is	
  provided	
  to	
  the	
  unit	
  by	
  the	
  research	
  leader.	
  	
  	
  
PAGE	
  -­‐	
  11	
  -­‐	
  
International	
  Institute	
  of	
  Tropical	
  Agriculture	
  (IITA)	
  
An	
  in-­‐house	
  survey	
  in	
  early	
  2013	
  revealed	
  that	
  most	
  research	
  data	
  is	
  stored	
  on	
  
individual	
   scientists	
   and	
   units	
   computers	
   in	
   various	
   different	
   software	
   and	
  
statistical	
   packages,	
   including	
   Excel,	
   SPSS,	
   Stata,	
   Access,	
   and	
   SQL.	
   Data	
   on	
  
research	
   projects/outputs	
   is	
   collected	
   at	
   an	
   ad	
   hoc	
   basis	
   (through	
  
questionnaires,	
  emails,	
   letters,	
   interviews,	
  group	
  discussions,	
  reviews	
  of	
  official	
  
reports).	
  The	
  data	
  is	
  produced	
  and	
  used	
  in	
  a	
  variety	
  of	
  formats,	
  including	
  digital,	
  
print	
   and	
   physical.	
   Most	
   of	
   research	
   data	
   are	
   still	
   very	
   fragmented.	
   IITA	
  
Biometrics	
   Unit	
   conducts	
   regular	
   training	
   and	
   backstopping	
   on	
   data	
   collection	
  
and	
  analysis	
  and	
  supports	
  genome	
  sequencing	
  projects,	
  specifically	
  in	
  generating	
  
the	
   NGS	
   (next	
   generation	
   sequencing)	
   data	
   for	
   gene-­‐phenotype	
   association	
  
studies.	
  IITA’s	
  GIS	
  unit	
  manages	
  a	
  geospatial	
  database	
  and	
  related	
  products.	
  	
  The	
  
survey	
   also	
   revealed	
   that	
   researchers	
  were	
   highly	
   supportive	
   of	
   having	
   a	
   data	
  
and	
  information	
  framework	
  as	
  they	
  felt	
  that	
  it	
  would	
  reduce	
  duplication	
  of	
  data	
  
collection	
  and	
  survey	
  fatigue	
  among	
  NARS	
  and	
  other	
  partners.	
  
IITA	
  has	
  taken	
  steps	
  to	
  improve	
  data	
  management.	
  One	
  of	
  the	
  first	
  successes	
  is	
  a	
  
relational	
   database	
   on	
   Cassava,	
   which	
   provides	
   cassava	
   breeders	
   and	
  
researchers	
  access	
   to	
  data	
  and	
   tools	
   in	
  a	
   centralized,	
  user-­‐friendly	
  and	
  reliable	
  
database.	
   IITA	
   is	
   looking	
   at	
   developing	
   other	
   crop	
   data	
   databases	
   using	
   the	
  
Cassavabase	
  as	
  a	
  model.	
  The	
  formulation	
  of	
  the	
  functional	
  requirements	
  for	
  this	
  
comprehensive	
   Crop	
   Breeding	
   Data	
   Management	
   Platform	
   based	
   on	
   breeders’	
  
needs	
   is	
   underway.	
   This	
   platform	
   would	
   include	
   the	
   management	
   of	
   crop	
  
information	
  and	
  the	
  development	
  of	
  applications	
  to	
  facilitate	
  breeding	
  processes	
  
and	
   agronomic	
   field	
   trials	
   data,	
   soil	
   science	
   data,	
   plant	
   health	
   information	
  
(pathology	
  and	
  entomology)	
  and	
  nutrition/post-­‐harvest	
  characteristics.	
  
IITA	
   is	
   also	
   envisioning	
   implementing	
   an	
   E-­‐Research	
   infrastructure	
   that	
   will	
  
support	
   research	
  and	
   that	
   enables	
   researchers	
   to	
  undertake	
  excellent	
   research	
  
and	
   deliver	
   innovation	
   outcomes,	
   provide	
   the	
   means	
   to	
   manipulate,	
   manage,	
  
share,	
   integrate	
   and	
   reuse	
   research	
   data,	
   and	
   enables	
   research	
   teams	
   to	
   share	
  
resources	
  and	
  work	
  together	
  more	
  effectively.	
  	
  
International	
  Maize	
  and	
  Wheat	
  Improvement	
  Center	
  (CIMMYT)	
  
CIMMYT	
   is	
  dealing	
  with	
  a	
  broad	
  variety	
  of	
  data	
  ranging	
   from	
  maize	
  and	
  wheat	
  
germplasm,	
   crop	
   data	
   at	
   the	
   field,	
   farm,	
   community,	
   country,	
   to	
   regional	
   and	
  
global	
   level,	
   socioeconomic	
   data.	
   Data	
   are	
   handled	
   by	
   several	
   units	
   within	
   or	
  
across	
  the	
  programs,	
  institution	
  and	
  partners.	
  CIMMYT	
  has	
  been	
  a	
  key	
  developer	
  
of	
   germplasm	
   related	
   data	
   management	
   platforms	
   including	
   Fieldbook,	
  
IMIS/IWIS,	
  and	
  the	
  IBP.	
  Data	
  volumes	
  have	
  increased	
  rapidly	
  and	
  with	
  that	
  the	
  
need	
   to	
   develop	
   tools	
   that	
   allow	
   the	
   manipulation	
   and	
   use	
   of	
   such	
   data	
   (eg	
  
genomics	
   applications;	
   decision	
   support	
   tools	
   for	
   sustainable	
   intensification	
  
approaches).	
  	
  In	
  2012	
  a	
  new	
  data	
  management	
  policy	
  that	
  came	
  into	
  effect	
  that	
  is	
  
to	
   accelerate	
  data	
   interchange	
   internally	
   and	
   externally.	
   	
   Six	
   data	
   coordinators	
  
were	
  hired	
  to	
  assist	
   in:	
  a)	
  establishing	
  data	
  standards,	
  documentation	
  and	
  data	
  
curation	
   processes,	
   b)	
   coordinate	
   receipt,	
   storage,	
   manipulation	
   and	
   quality	
  
control	
  of	
  field	
  and	
  germplasm	
  related,	
  c)	
  participate	
  in	
  the	
  design,	
  development	
  
and	
   population	
   of	
   versatile	
   institutional	
   databases/repositories,	
   interfaces	
   and	
  
output	
   tools,	
   d)	
   introduce	
   new	
   informatics	
   tools	
   to	
   staff	
   and	
   collaborators,	
  
provide	
   on-­‐the-­‐job	
   training,	
   and	
   report	
   back	
   user	
   requirements	
   to	
   CIMMYT	
  
software	
  engineers,	
  e)	
  manage	
  and	
  document	
  the	
  dissemination	
  of	
  data	
  (raw	
  and	
  
analyzed).	
  
PAGE	
  -­‐	
  12	
  -­‐	
  
There	
   are	
   currently	
   three	
  data	
   coordinators	
   in	
   the	
  Genetic	
  Resources	
  Program	
  
(GRP)	
   who	
   work	
   for	
   the	
   Global	
   Maize	
   Program	
   (GMP)	
   and	
   the	
   Global	
   Wheat	
  
Program	
   (GWP)	
   handling	
   both	
  molecular	
   data	
   and	
   breeder’s	
   trial	
   and	
   nursery	
  
data	
   (phenotypic).	
   The	
   breeder’s	
   nursery	
   and	
   trial	
   data	
   mainly	
   consists	
   of	
  
phenotypic	
   data	
   collected	
   in	
   the	
   field	
   during	
   the	
   crop	
   cycle	
   and	
   can	
   be	
  
complemented	
  by	
  quality	
  and	
  nutritional	
  traits	
  (such	
  as	
  nutritional	
  value,	
  forage	
  
traits,	
   baking	
   quality	
   or	
   other	
   food	
   processing	
   relevant	
   traits)	
   or	
   molecular	
  
information	
   at	
   variable	
   density	
   (few	
   markers)	
   to	
   several	
   genotype-­‐by-­‐
sequencing	
   information	
   across	
   significant	
   parts	
   of	
   the	
   genome.	
   The	
   socio	
  
economics	
   program	
   (SEP)	
   and	
   the	
   global	
   conservation	
   agriculture	
   program	
  
(GCAP)	
   share	
   three	
   data	
   coordinators	
   who	
   work	
   in	
   the	
   regions	
   (Africa,	
   Asia,	
  
Latin	
  America	
  as	
  well	
   as	
  globally)	
  on	
  both	
  socio	
  economic	
  data	
  and	
  agronomic	
  
trial	
  data,	
  their	
  standardizations	
  and	
  platforms.	
  
The	
  GIS	
  unit	
  collects	
  spatial	
  and	
  meteorological	
  data	
  on	
  global,	
  national	
  and	
  sub	
  
national	
  scales	
  as	
  well	
  as	
  statistical	
  data	
  related	
  to	
  maize	
  and	
  wheat	
  production	
  
in	
   all	
   countries	
   producing	
   these	
   two	
   commodities	
   which	
   is	
   then	
   converted	
   to	
  
spatial	
   data	
   and	
   as	
   geo	
   referenced	
   data	
   collected	
   and	
   generated	
   in	
   specific	
  
projects.	
  	
  	
  	
  	
  	
  	
  
CIMMYT	
  knowledge	
  management	
  is	
  co-­‐chairing	
  the	
  Wheat	
  Data	
  Interoperability	
  
working	
  group	
  which	
  aims	
  to	
  coordinate	
  worldwide	
  research	
  efforts	
  in	
  the	
  fields	
  
of	
  wheat	
  genetics,	
  genomics,	
  physiology,	
  breeding	
  and	
  agronomy.	
  This	
  part	
  of	
  the	
  
Research	
  Data	
  Alliance	
  initiative.	
  
International	
  Rice	
  Research	
  Institute	
  (IRRI)	
  
IRRI	
   has	
   various	
   research	
   support	
   groups	
   with	
   different	
   roles	
   and	
  
responsibilities.	
   The	
   Research	
   Data	
   Management	
   (RDM)	
   group	
  
conducts	
  regular	
  training	
  on	
  good	
  practices	
  in	
  managing	
  research	
  data	
  to	
  ensure	
  
that	
   data	
   are	
   well	
   documented	
   (metadata)	
   and	
   organized	
   systematically	
  
(file	
  repositories).	
  	
  Areas	
  covered	
  include:	
  Research	
  data	
  planning;	
  Research	
  data	
  
collection,	
   authentication	
   and	
   storage;	
   Data	
   backup	
   and	
   security;	
   and	
   Data	
  
archival	
  and	
  sharing.	
  RDM	
  also	
  assist	
  research	
  staff	
  in	
  implementing	
  these	
  good	
  
practices	
   based	
   on	
   RDM	
   policy	
   and	
   IP	
   Policy.	
   The	
   Biometrics	
   and	
   Breeding	
  
Informatics	
   group	
   conducts	
  regular	
  training	
   on	
   experimental	
   designs	
   and	
  
statistical	
  data	
  analysis	
  as	
  well	
  as	
  the	
  use	
  of	
  plant	
  breeding	
  tools	
  developed	
  in-­‐
house.	
   The	
   Bioinformatics	
   group	
   conducts	
  training	
   on	
   Bioinformatics	
   and	
  
supports	
  research	
  groups	
  by	
  providing	
  tools	
  for	
  SNP	
  analysis.	
  
All	
   instruments	
   and	
   lab	
   equipment	
   are	
   ISO	
   certified	
   and	
   in-­‐house	
   verification	
  
performed	
   regularly,	
   to	
   ensure	
   accuracy	
   and	
   traceability	
   of	
   research	
   data	
  
collected.	
   Additional	
   flagship	
   databases	
   that	
   are	
   maintained	
   by	
   research	
   units	
  
are	
   ICIS-­‐IRIS	
   for	
   germplasm	
   data,	
  World	
   Rice	
   Statistics	
   and	
   Household	
   survey	
  
database	
   for	
  Social	
  Science	
  unit,	
  Long	
  Term	
  Continuous	
  Cropping	
  Experiments,	
  
and	
  Climate	
  data	
  for	
  Climate	
  Unit.	
  	
  IRRI	
  also	
  maintains	
  the	
  Rice	
  Knowledge	
  bank,	
  
which	
   is	
   the	
   world’s	
   leading	
   repository	
   of	
   extension	
   and	
   training	
   materials	
  
related	
  to	
  rice	
  production.	
  	
  	
  
International	
  Livestock	
  Research	
  Institute	
  (ILRI)	
  
The	
   ILRI	
   Research	
   Methods	
   Group	
   consists	
   of:	
   2	
   statisticians,	
   5	
   systems	
  
designers	
   /	
  managers,	
   2	
   systems	
   administrators	
   (high	
   performance	
   computing	
  
systems)	
  and	
  2	
  GIS	
  data	
  managers	
  /	
  analysts.	
  In	
  addition	
  we	
  have	
  a	
  large	
  number	
  
PAGE	
  -­‐	
  13	
  -­‐	
  
of	
  GIS	
  analysts	
  sat	
  in	
  research	
  programs	
  and	
  a	
  few	
  database	
  managers	
  and	
  data	
  
systems	
   type	
   people.	
   There	
   is	
   also	
   a	
   bioinformatics	
   group	
   which	
   sits	
   in	
   the	
  
Directorate	
  of	
  Biosciences	
  (BECA	
  &	
  Animal	
  Biosciences).	
  Management	
  of	
  data	
  is	
  
mainly	
   at	
   the	
   project	
   level	
   although	
   for	
   large	
   projects	
   shared	
   SQL	
   servers	
   are	
  
used.	
  With	
  the	
  advent	
  of	
  the	
  CGIAR	
  Research	
  Programs	
  (CRPs)	
  ILRI	
  is	
  currently	
  
establishing	
  a	
  data	
  portal	
  /	
  platform	
  for	
  both	
  Institute	
  and	
  the	
  CRP	
  on	
  Livestock	
  
and	
   Fish	
   data	
   publishing.	
   Initially	
   the	
   data	
   portal	
  will	
   provide	
   cataloguing	
   and	
  
access	
  to	
  data	
  and	
  in	
  later	
  development	
  a	
  module	
  for	
  monitoring	
  and	
  evaluation	
  
will	
   allow	
   for	
  meta-­‐analysis	
   across	
  projects,	
   for	
  key	
   indicators	
   (e.g.	
   IDO’s).	
  The	
  
open-­‐source	
   platform	
   (CKAN)	
   will	
   also	
   allow	
   inter-­‐operability	
   and	
  
communication	
  with	
  ILRI’s	
  knowledge	
  management	
  system	
  (DSpace	
  –	
  Mahider)	
  
and	
   other	
   platforms.	
   A	
   revised	
   data	
   management	
   policy	
   aligned	
   to	
   the	
   CGIAR	
  
Open-­‐access	
  policy,	
  the	
  CGIAR	
  Management	
  of	
  Intellectual	
  Assets	
  (IA)	
  policy	
  and	
  
various	
  CRP	
  policies	
  in	
  development	
  is	
  currently	
  being	
  finalized	
  with	
  detailed	
  a	
  
detailed	
   implementation	
  plan	
  providing	
   ILRI	
  staff	
  and	
  partners	
  will	
  options	
   for	
  
collecting,	
  managing	
  and	
  sharing	
  their	
  data.	
  
International	
  Water	
  Management	
  Institute	
  (IWMI)	
  
IWMI’s	
   Information	
   and	
   Knowledge	
   Group	
   (IKG)	
   provides	
   knowledge	
  
management	
   support,	
   connecting	
   IWMI’s	
   knowledge	
   outputs	
   through	
   the	
  
integrated	
   Institution-­‐wide	
   search	
   facility,	
   “Poodle”	
   that	
   allows	
   for	
   a	
  
comprehensive	
   search	
   across	
   peer	
   reviewed	
   and	
   non-­‐peer	
   reviewed	
   IWMI	
  
research	
  outputs.	
  Amongst	
  many	
  other	
  responsibilities	
  IKG	
  provides	
  publication	
  
support	
   and	
  manages	
   IWMI’s	
   internal	
   publications	
   such	
   as	
   the	
   IWMI	
   working	
  
paper	
   and	
   IWMI	
   research	
   report	
   series.	
   IWMI	
   currently	
   has	
   no	
   separate	
   Data	
  
Management	
   Unit,	
   but	
   its	
   Research	
   data	
  Management	
   Policy	
   stipulates	
   that	
   all	
  
data	
   must	
   be	
   archived	
   in	
   IWMI’s	
   central	
   repository	
   (WDP)	
   with	
   standard	
  
metadata	
   as	
   early	
   as	
   possible	
   after	
   collection	
   and	
   processing	
  with	
   appropriate	
  
access	
  rights.	
  Project	
   leaders	
  are	
  held	
  responsible	
   for	
  ensuring	
   implementation	
  
of	
  data	
  management	
  policy	
  at	
  project	
  level	
  while	
  IWMI	
  theme	
  leaders	
  ensure	
  that	
  
all	
   projects	
   under	
   the	
   themes	
   are	
   complying	
   with	
   the	
   policy.	
   IWMI’s	
  
Implementation	
   Framework	
   for	
   research	
   data	
   management	
   follows	
   that	
  
developed	
  and	
  used	
  by	
  ICRAF	
  (2008)	
  (IWMI	
  2011).	
  
IWMI	
   has	
   a	
   dedicated	
  Geo-­‐informatics	
  Unit	
   (GRandD)	
  which	
   is	
   responsible	
   for	
  
developing	
   data	
   standards,	
   protocol	
   and	
   training	
   to	
   researchers	
   and	
   research	
  
assistants	
   to	
   use	
   these	
   standards.	
   The	
   unit	
   provides	
   research	
   support	
   for	
  
database	
  design,	
  data	
  organization,	
  data	
  manipulation,	
  metadata	
  preparation	
  and	
  
data	
   archiving.	
   The	
   GRandD	
   unit	
   coordinates	
  with	
   projects	
   for	
   uploading	
   data	
  
into	
  the	
  central	
  repository,	
  making	
  them	
  available	
  to	
  appropriate	
  users	
  through	
  
its	
  Water	
   Data	
   Portal,	
   an	
   integrated	
   portal	
   for	
   consulting	
   and	
   accessing	
   IWMI	
  
research	
  data.	
  
World	
  Agroforestry	
  Centre	
  (ICRAF)	
  
ICRAF	
  has	
  a	
  Research	
  Methods	
  Group	
  as	
  part	
  of	
   their	
  global	
  support	
  units.	
  The	
  
group	
   provides	
   services	
   all	
   along	
   the	
   research	
   cycle	
   from	
   development	
   of	
  
conceptual	
  frameworks	
  consisting	
  of	
  problem	
  definition,	
  hypothesis,	
  models	
  and	
  
research	
   questions,	
   well-­‐documented	
   research	
   designs	
   and	
   methods,	
   data	
  
management	
  and	
  curation	
  and	
  statistical	
  analysis.	
  At	
  present	
  the	
  group	
  consists	
  
of	
  three	
  statisticians	
  and	
  three	
  data	
  base	
  managers	
  at	
  Headquarter.	
  	
  The	
  group	
  is	
  
PAGE	
  -­‐	
  14	
  -­‐	
  
further	
  supported	
  through	
  3	
  regional	
  data	
  managers,	
  with	
  further	
  recruitments	
  
in	
  the	
  regions	
  line	
  up	
  for	
  2014.	
  	
  
In	
   2012	
   about	
   65%	
  of	
   the	
  work	
  was	
   related	
   to	
   data	
  management	
   and	
   35%	
   to	
  
research	
  design	
  and	
  data	
  analysis.	
  	
  
Since	
  November	
  2012	
  ICRAF	
  has	
  a	
  research	
  data	
  management	
  policy	
  that	
  makes	
  
it	
  mandatory	
   for	
   scientists	
   to	
  provide	
  open	
  access	
   to	
   all	
   relevant	
  primary	
  data	
  
that	
   are	
   accompanying	
   their	
   scientific	
   publication.	
   Projects	
   are	
   responsible	
   to	
  
ensure	
   that	
   research	
   data	
   is	
   described	
   by	
   appropriated	
   Metadata	
   throughout	
  
their	
   lifecycle	
   and	
   are	
   required	
   to	
   have	
   all	
   their	
   datasets	
   submitted	
   to	
   the	
  
repository	
  upon	
  closure.	
  	
  Metadata	
  are	
  move	
  to	
  the	
  open	
  domain	
  as	
  soon	
  as	
  they	
  
have	
   been	
   compiled,	
   the	
   actual	
   data	
   sets	
   are	
   only	
   released	
   once	
   the	
   scientists	
  
give	
  their	
  approval.	
  
ICRAF	
  is	
  using	
  Dataverse	
  as	
  its	
  data	
  repository,	
  which	
  was	
  released	
  on	
  October	
  
2011	
  2011.	
   	
   As	
   of	
   September	
  2013	
  we	
  have	
  96	
   studies	
   in	
   our	
  Dataverse.	
   	
   The	
  
usage	
   is	
   still	
   largely	
   internal,	
   especially	
   for	
   ongoing	
   projects.	
   From	
   our	
  
experience	
   in	
   the	
   last	
   20	
   month	
   we	
   find	
   that	
   projects	
   are	
   more	
   likely	
   to	
  
implement	
  the	
  policy	
  if	
  they	
  have	
  access	
  to	
  a	
  data	
  manager	
  in	
  the	
  same	
  office.	
  To	
  
share	
  datasets	
  and	
  metadata	
  with	
  partner	
  organization	
  ICRAFs	
  prime	
  platform	
  is	
  
its	
   GeoPortal.	
   It	
   is	
   primarily	
   designed	
   to	
   provide	
   researchers	
  with	
   secure	
   data	
  
storage,	
  sharing	
  and	
  visualization	
  options	
  through	
  its	
  Web	
  Mapping	
  Application.	
  
Ultimately,	
  the	
  GeoPortal	
  will	
  be	
  a	
  full-­‐fledged	
  online	
  GIS	
  tool	
  with	
  a	
  number	
  of	
  
features	
   for	
   visualization,	
   data	
   management	
   and	
   spatial	
   modeling.	
   Being	
  
designed	
  using	
  exclusively	
  open	
  source	
  platforms	
  and	
  tools	
  it	
  is	
  being	
  used	
  by	
  an	
  
increasing	
   number	
   of	
   CGIAR	
   scientists,	
   as	
   well	
   as	
   scientists	
   from	
   partnering	
  
institutions.	
  
WorldFish	
  
At	
   WorldFish,	
   the	
   Research	
   Data	
   Management	
   Project	
   (RDMP),	
   together	
   with	
  
three	
   research	
   support	
   people,	
   is	
   adding	
   value	
   to	
   data	
   collected	
   by	
   WF	
   staff	
  
across	
  all	
  our	
  offices.	
  The	
  goal	
  of	
   the	
   initiative	
   is	
   to	
  make	
  all	
  data	
  produced	
  by	
  
Worldfish	
  	
  available	
  firstly	
  within	
  Worldfish	
  and	
  perhaps	
  more	
  widely	
  given	
  the	
  
necessary	
   permissions.	
   Project	
   Leaders	
   are	
   required	
   to	
   submit	
   well	
   organized	
  
and	
  well-­‐described	
  data	
   files	
  after	
   their	
  projects	
   terminate.	
   	
  The	
  RDMP	
  team	
  is	
  
trying	
  to	
  collate	
  these	
  individual	
  data	
  sets	
  into	
  a	
  relational	
  database	
  that	
  allows	
  
global	
   analysis.	
  WFA	
   also	
   just	
   set	
   up	
   a	
   GIS	
   and	
   data	
   helpdesk	
   for	
   staff	
   to	
   use.	
  
‘Clients’	
   can	
   request	
   specific	
   data,	
   graphs,	
   maps	
   and	
   other	
   data	
   products.	
  
Additionally	
  the	
  team	
  also	
  manages	
  ReefBase	
  and	
  The	
  Coral	
  Triangle	
  Atlas	
  which	
  
incorporate	
  online	
  geo-­‐referenced	
  data.	
  
Success	
  stories	
  
Some	
  example	
  success	
  stories	
  of	
  CGIAR	
  open	
  data:	
  
AgTrials	
  
AgTrials:	
   http://www.agtrials.org/	
   is	
   an	
   information	
   portal	
   developed	
   by	
   the	
  
CGIAR	
   Research	
   Program	
   on	
   Climate	
   Change,	
   Agriculture	
   and	
   Food	
   Security	
  
(CCAFS)	
  which	
  provides	
  access	
  to	
  a	
  database	
  on	
  the	
  performance	
  of	
  agricultural	
  
technologies	
   at	
   sites	
   across	
   the	
   developing	
   world.	
   It	
   builds	
   on	
   decades	
   of	
  
evaluation	
  trials,	
  mostly	
  of	
  varieties,	
  but	
  includes	
  any	
  agricultural	
  technology	
  for	
  
PAGE	
  -­‐	
  15	
  -­‐	
  
developing	
  world	
  farmers.	
  This	
  project	
  will	
  standardize	
  data	
  and	
  information	
  to	
  
the	
   benefit	
   of	
   climate	
   change	
   analyses,	
   future	
   multi-­‐environment	
   trials	
   and	
  
research	
   and	
   development	
   in	
   international	
   agriculture.	
  With	
   the	
   interface	
   you	
  
can	
   share	
   data	
   and	
   information	
   on	
   evaluations	
   of	
   agricultural	
   technology;	
  
acquire	
   agricultural	
   evaluation	
   data	
   sets	
   for	
   your	
   own	
   research;	
   explore	
   the	
  
geographic	
  dimensions	
  of	
  agricultural	
  evaluation.	
  
ASTI	
  
ASTI:	
   Agricultural	
   Science	
   and	
   Technology	
   Indicators	
   www.asti.cgiar.org	
   is	
   a	
  
database	
   that	
   provides	
   up-­‐to-­‐date	
   do	
   quantitative	
   and	
   qualitative	
   data	
   on	
  
investment,	
   capacity,	
   and	
   institutional	
   trends	
   in	
   agricultural	
   research	
   and	
  
development.	
  The	
  dataset	
  are	
  collected	
  and	
  updated	
  annually.	
  ASTI	
  data	
  are	
  used	
  
in	
  informing	
  policy	
  formulation	
  and	
  decision	
  making	
  in	
  many	
  countries	
  of	
  Africa	
  
and	
  South	
  Asia.	
  	
  
Ethiopia	
  Rural	
  Household	
  Surveys	
  
Ethiopia	
  Rural	
  Household	
  Surveys	
  (Hoddinott	
  and	
  Yohannes	
  (2011).	
  This	
  dataset	
  
has	
  resulted	
  in	
  many	
  publications	
  since	
  its	
  release	
  in	
  2011.	
  This	
  dataset	
  has	
  been	
  
cited	
  at	
  least	
  10	
  times.	
   	
  Furthermore,	
  the	
  publications	
  produced	
  with	
  the	
  use	
  of	
  
this	
   dataset	
   have	
   also	
   informed	
   policy	
  making	
   in	
   Ethiopia.	
  We	
   receive	
   a	
   lot	
   of	
  
personal	
  request	
  to	
  provide	
  this	
  dataset	
  as	
  well	
  in	
  addition	
  to	
  web	
  downloads.	
  
Chronic	
  Poverty	
  and	
  Long	
  Term	
  Impact	
  Study	
  in	
  Bangladesh	
  
Chronic	
  Poverty	
   and	
  Long	
  Term	
   Impact	
   Study	
   in	
  Bangladesh	
   (Quisumbing	
   and	
  
Baulch,	
   2010):	
   According	
   to	
   Thomson	
   Reuter’s	
   Data	
   Citation	
   Index,	
   this	
  
particular	
  dataset	
   has	
  been	
   cited	
  more	
   than	
  24	
   times	
   since	
   its	
   release	
   in	
  2010	
  
and	
  downloaded	
  3479	
  times.	
  
Land	
  Degradation	
  Surveillance	
  Framework	
  
Land	
  Degradation	
  Surveillance	
  Framework	
  (LDSF),	
  
http://gsl.worldagroforestry.org/?q=node/239,	
  is	
  a	
  sampling	
  protocol	
  designed	
  
around	
  a	
  spatially	
  stratified,	
  randomized	
  sampling	
  design,	
  to	
  provide	
  a	
  
biophysical	
  baseline	
  at	
  landscape	
  level.	
  The	
  LDSF	
  was	
  developed	
  at	
  the	
  World	
  
Agroforestry	
  Centre	
  for	
  landscape	
  level	
  assessments	
  and	
  studies	
  of	
  carbon	
  
dynamics,	
  vegetation	
  changes,	
  soil	
  functional	
  properties	
  and	
  soil	
  hydrological	
  
properties.	
  The	
  LDSF	
  has	
  been	
  implemented	
  in	
  more	
  than	
  20	
  countries	
  in	
  Africa	
  
to	
  date,	
  including	
  the	
  CIAT-­‐led	
  Africa	
  Soil	
  Information	
  Service	
  (AfSIS)	
  project.	
  
The	
  methodology	
  has	
  been	
  shown	
  to	
  be	
  appropriate	
  for	
  studies	
  of	
  land	
  health	
  
and	
  land	
  degradation	
  risk,	
  as	
  well	
  as	
  for	
  assessing	
  soil	
  organic	
  carbon	
  dynamics	
  
in	
  rangeland	
  systems.	
  
Poverty	
  Environmental	
  Network	
  
Poverty	
  Environmental	
  Network	
  PEN.	
  Launched	
  in	
  2004,	
  PEN,	
  
http://www.cifor.org/pen,	
  is	
  the	
  largest	
  and	
  most	
  comprehensive	
  global	
  analysis	
  
of	
  tropical	
  forests	
  and	
  poverty.	
  Its	
  database	
  contains	
  survey	
  data	
  on	
  8000+	
  
households	
  in	
  40+	
  study	
  sites	
  in	
  25	
  developing	
  countries.	
  At	
  the	
  core	
  of	
  PEN	
  is	
  
comparative,	
  detailed	
  socio-­‐economic	
  data	
  that	
  was	
  collected	
  quarterly	
  at	
  the	
  
household	
  and	
  village	
  level	
  by	
  50+	
  research	
  partners	
  using	
  standardized	
  
definitions,	
  questionnaires	
  and	
  methods.	
  
PAGE	
  -­‐	
  16	
  -­‐	
  
Reefbase,	
  and	
  the	
  CT-­‐Atlas	
  
Reefbase,	
  and	
  the	
  CT-­‐Atlas	
  (http://www.reefbase.org/main.aspx;	
  
http://ctatlas.reefbase.org/)	
  are	
  online	
  GIS	
  database	
  systems	
  developed	
  by	
  
WorldFish	
  and	
  partners.	
  This	
  year	
  they	
  have	
  been	
  recognized	
  by	
  the	
  Thematic	
  
Working	
  Groups	
  of	
  the	
  Coral	
  Triangle	
  Initiative	
  as	
  their	
  official	
  data	
  storage	
  and	
  
retrieval	
  tools.	
  	
  These	
  databases	
  are	
  improving	
  the	
  regional	
  coordination	
  of	
  
conservation	
  and	
  management	
  activities	
  in	
  the	
  Coral	
  Triangle	
  region.	
  The	
  CT-­‐
Atlas	
  will	
  be	
  maintained	
  in	
  future	
  with	
  funding	
  from	
  CCAFS,	
  ADB	
  and	
  the	
  IAEA	
  
(International	
  Atomic	
  Energy	
  Agency)	
  and	
  will	
  start	
  to	
  include	
  more	
  socio-­‐
economic	
  datasets,	
  and	
  be	
  directed	
  more	
  towards	
  the	
  examination	
  of	
  food	
  
security	
  issues.	
  	
  
Longitudinal	
  Village	
  Level	
  Studies	
  
The	
   longitudinal	
   Village	
   Level	
   Studies	
   (VDSA),	
   	
   http://vdsa.icrisat.ac.in/vdsa-­‐
vls.htm,	
  	
  of	
  ICRISAT	
  have	
  for	
  over	
  three	
  decades	
  provided	
  profound	
  insights	
  into	
  
the	
  social	
  and	
  economic	
  changes	
   in	
  the	
  village	
  and	
  household	
  economies	
   in	
  the	
  
semi-­‐arid	
  tropics	
  of	
  Asia	
  and	
  Africa.	
  Over	
  150	
  research	
  papers	
  and	
  more	
  than	
  40	
  
doctoral	
  dissertations	
  have	
  been	
  based	
  on	
  empirical	
  analysis	
  of	
  VLS	
  data	
  in	
  the	
  
semi-­‐arid	
   tropics	
   of	
   India	
   and	
  West	
   Africa.	
   A	
   recent	
   search	
   in	
   Google	
   scholar	
  
shows	
  that	
  this	
  body	
  of	
  work	
  has	
  generated	
  over	
  10,000	
  citations.	
  
Cassavabase	
  
Cassavabase:	
   The	
   Next	
   Generation	
   Cassava	
   Breeding	
   (NEXTGEN	
   Cassava)	
  
project,	
   implemented	
   in	
   collaboration	
  with	
  Cornell	
  University,	
  has	
  developed	
  a	
  
database	
   which	
   gives	
   The	
   database	
   (www.cassavabase.org)	
   contains	
   Genomic	
  
Selection	
  algorithms	
  and	
  analysis	
   capacity,	
   a	
   cassava	
  genome	
  browser,	
   cassava	
  
ontology	
  tools,	
  phenotyping	
  tools,	
  and	
  social	
  networking.	
  Tools	
  are	
  developed	
  on	
  
Cassavabase	
   that	
   improve	
   partner	
   breeding	
   program	
   information	
   tracking,	
  
streamline	
  management	
  of	
  genotypic	
  and	
  phenotypic	
  data,	
  and	
  pipeline	
  that	
  data	
  
through	
  Genomic	
  Selection	
  prediction	
  analyses.	
  By	
  the	
  project	
  end,	
  Cassavabase	
  
will	
  be	
  fully	
  hosted	
  at	
  IITA,	
  providing	
  a	
  "one-­‐stop	
  shop"	
  for	
  cassava	
  researchers	
  
and	
  breeders	
  worldwide.	
  
CIAT	
  Geonetwork	
  
CIAT	
  Geonetwork	
  
(http://gisweb.ciat.cgiar.org:8080/geonetwork/srv/en/about),	
  the	
  SRTM	
  digital	
  
elevation	
  data	
  (http://srtm.csi.cgiar.org/)	
  and	
  Worldclim	
  
(http://www.worldclim.org)	
  are	
  great	
  examples.	
  Some	
  of	
  these	
  datasets	
  and	
  
associated	
  information	
  has	
  been	
  downloaded	
  by	
  several	
  thousand	
  users.	
  The	
  
research	
  paper	
  published	
  in	
  2005	
  (CIAT	
  and	
  Bioversity)	
  by	
  Hijmans	
  et	
  al	
  (2005)	
  
very	
  high	
  resolution	
  interpolated	
  climate	
  surfaces	
  for	
  global	
  land	
  areas	
  
(Worldclim)’’	
  is	
  one	
  of	
  the	
  most	
  cited	
  article	
  in	
  CGIAR	
  history.	
  
Intergenebank	
  Potato	
  Database	
  	
  
The	
  Intergenebank	
  Potato	
  Database	
  (IPD),	
  a	
  database	
  to	
  cross-­‐reference	
  potato	
  
collections	
  and	
  share	
  passport	
  and	
  evaluation	
  data,	
  since	
  early	
  1995.	
  	
  
PAGE	
  -­‐	
  17	
  -­‐	
  
SINGER	
  
(Systems-­‐Wide	
   Information	
   Network	
   for	
   Genetic	
   Resources):	
   is	
   a	
   genetic	
  
resources	
  information	
  exchange	
  network	
  that	
  provides	
  access	
  to	
  information	
  on	
  
the	
   collections	
   of	
   genetic	
   resources.	
   Established	
   in	
   the	
   1990ies	
   the	
   collection	
  
now	
  comprise	
  over	
  half	
  a	
  million	
  samples	
  of	
  crop,	
  forage	
  and	
  tree	
  germplasm	
  of	
  
major	
  importance	
  for	
  food	
  and	
  agriculture.	
  
Barriers	
  to	
  mainstreaming	
  data	
  sharing	
  
Sensitive	
  data	
  &	
  data	
  confidentiality	
  
Data	
  sets	
  often	
  have	
  personal	
   information	
  about	
  households	
  or	
   informants	
  that	
  
they	
  do	
  not	
  expect	
  to	
  be	
  made	
  public	
  and	
  that	
  would	
  be	
  unethical	
  to	
  make	
  public,	
  
given	
   the	
   trust	
   and	
   rapport	
   between	
   researcher	
   and	
   ‘informant’.	
   	
   Sensitive	
  
research	
   data	
   include	
   data	
   on	
   illegal	
   activities,	
   corruption,	
   land	
   tenure	
   and	
  
conflicts,	
   controversial	
   governmental	
   policies,	
   the	
   location	
   of	
   wild	
   and	
  
domesticated	
   plant	
   genetic	
   resources	
   with	
   particularly	
   valuable	
   traits.	
   Several	
  
countries	
  have	
  some	
  form	
  of	
  data	
  protection	
   laws	
  to	
  regulate	
  the	
  processing	
  of	
  
information	
  relating	
  to	
  individuals	
  and	
  their	
  traditional	
  knowledge,	
  including	
  the	
  
obtaining,	
   holding,	
   use	
   or	
   disclosure	
   of	
   such	
   information.	
   Very	
   few	
   CGIAR	
  
Centers	
  have	
  ethical	
   review	
  boards	
  or	
  committees	
   to	
  ensure	
   that	
  data	
  used	
   for	
  
human	
  or	
  behavioral	
   research	
  complies	
  with	
  ethical	
   review	
  standards	
  and	
   that	
  
research	
   subjects	
   or	
   participants	
   are	
   protected.	
   While	
   there	
   are	
   a	
   number	
   of	
  
standard	
   procedures	
   to	
   anonymise	
   data	
   sets,	
   such	
   as	
   removing	
   names,	
  
addresses,	
   and	
   contact	
   information	
   or	
   encryption	
   of	
   location	
   data	
   (GPS)	
   it	
  
creates	
   a	
   secondary	
   problem	
   of	
   managing	
   different	
   data	
   sets:	
   one	
   complete	
  
version	
  for	
  internal	
  use	
  and	
  the	
  second	
  autonomous	
  data	
  set	
  for	
  public	
  use.	
  	
  
Diverse	
  data	
  sets	
  and	
  backlog	
  
The	
   interdisciplinary	
   nature	
   of	
   the	
   research	
   conducted	
   within	
   each	
   of	
   the	
  
Centers	
  results	
  in	
  a	
  vast	
  variety	
  of	
  data	
  formats	
  and	
  types.	
  Automated	
  workflows	
  
for	
   data	
   verification,	
   cleaning	
   and	
   aggregation	
   need	
   to	
   be	
   customized	
   for	
   each	
  
project,	
   resulting	
   in	
   high	
   demand	
   on	
   research	
   support	
   staff	
   time.	
   In	
   addition	
  
global	
   analyses	
   across	
   data	
   sets	
   from	
   different	
   projects	
   requires	
   relational	
  
databases	
   which	
   are	
   difficult	
   to	
   realize	
   if	
   data	
   collections	
   have	
   not	
   been	
  
standardized.	
  While	
  across	
  the	
  Centers	
  there	
  are	
  various	
  affords	
  to	
  identify	
  “key”	
  
indicators	
  that	
  should	
  be	
  collected	
  by	
  each	
  project	
  or	
  to	
  introduce	
  standardized	
  
sampling	
   designs	
   and	
   survey	
   modules	
   scientists	
   feel	
   that	
   scientific	
   creativity	
  
should	
  not	
  be	
  pressed	
  into	
  rigid	
  protocols.	
  	
  In	
  addition,	
  each	
  centre	
  has	
  a	
  legacy	
  
of	
  high	
  value	
  data	
  sets	
  that	
  have	
  not	
  been	
  fully	
  curated.	
  To	
  find	
  sufficient	
  budget	
  
to	
   update	
   these	
   datasets	
   and	
   to	
   make	
   them	
   available	
   within	
   and	
   outside	
   the	
  
centre	
  is	
  a	
  problem.	
  	
  
Data	
  ownership	
  and	
  recognition	
  of	
  data	
  authors	
  	
  
Data	
   cannot	
   easily	
   be	
   protected	
   against	
   copying	
   that	
   work,	
   or	
   reproducing	
   it	
  
without	
   authorization	
   or	
   attribution.	
   Copyright	
   applies	
   not	
   to	
   the	
   facts	
   or	
   the	
  
information	
   itself,	
   but	
   to	
   the	
   particular	
   way	
   the	
   facts	
   or	
   information	
   are	
  
presented	
   in	
   the	
   dataset	
   or	
   database.	
   As	
   such	
   a	
   database	
   can	
   be	
   protected	
   by	
  
copyright,	
   but	
   only	
   the	
   database	
  model	
   and	
   the	
   data	
   entry	
   and	
   output	
   not	
   the	
  
PAGE	
  -­‐	
  18	
  -­‐	
  
actual	
   numbers	
   or	
   names	
   in	
   the	
   database.	
   	
   Data	
   sharing	
   websites	
   such	
   as	
  
DataCite	
  or	
  Dataverse	
  assign	
  a	
  persistent	
  authorship	
   identifier	
   (URI),	
   such	
  as	
  a	
  
digital	
   object	
   identifier	
   (DOI)	
   or	
   handle	
   to	
   data	
   sets	
   and	
   have	
   specific	
   user	
  
agreements.	
  However,	
  while	
  these	
  might	
  be	
  legally	
  binding	
  in	
  some	
  countries	
  in	
  
it	
  not	
  clear	
  how	
  scientist	
  or	
  centers	
  can	
  take	
  legal	
  actions	
  against	
  misuse.	
  While	
  
the	
  ownership	
  of	
  the	
  data	
  and	
  the	
  right	
  to	
  reproduce	
  the	
  work	
  usually	
  belongs	
  to	
  
the	
   centers,	
   scientists	
   are	
   given	
   authorship	
   rights.	
   Unlike	
   scientific	
   publication	
  
there	
  are	
  no	
  standard	
  guidelines	
  regarding	
  data	
  authorship.	
  	
  Centers	
  and	
  project	
  
managers	
  need	
  to	
  ensure	
  that	
  both	
  technical	
  as	
  well	
  as	
  scientific	
  staff	
  are	
  given	
  
the	
   deserved	
   credit	
   for	
   their	
   work,	
   thus	
   all	
   people	
   that	
   have	
   substantially	
  
contributed	
  to	
  the	
  creation	
  of	
  the	
  datasets	
  should	
  be	
  data	
  authors.	
  	
  
For	
   projects	
   that	
   consist	
   of	
   multiple	
   datasets	
   produced	
   by	
   different	
   teams	
   of	
  
scientists,	
  decisions	
  need	
  to	
  be	
  made	
  about	
  assigning	
  the	
  authorship.	
  Assigning	
  
the	
   same	
  authorship	
   (the	
   same	
  persistent	
   identifier)	
   to	
   all	
   project	
   related	
  data	
  
ensures	
   the	
   coherence	
   of	
   the	
   datasets,	
   but	
   it	
   does	
   not	
   allow	
   differentiating	
  
between	
   the	
   different	
   contributions	
   of	
   scientists,	
   which	
   is	
   problematic	
   with	
  
respect	
   to	
   accountability.	
   The	
   same	
   issues	
   arise	
   for	
   dynamic	
   datasets	
   such	
   as	
  
R&D	
   databases	
   and	
   trial	
   data	
   that	
   data.	
   	
   Other	
   organizations	
   with	
   a	
   stronger	
  
mandate	
  to	
  produce	
  global	
  data	
  sets	
  as	
  public	
  goods	
  such	
  as	
  the	
  OECD	
  and	
  FAO	
  
attributing	
   institution/program	
   as	
   the	
   data	
   authors	
   for	
   dynamic	
   datasets,	
  
individual	
  contribution	
  are	
  recognized	
  within	
   list	
  of	
   contributors	
  or	
  mentioned	
  
in	
  acknowledgement.	
  Assigning	
  authorship	
   for	
  value	
  –added	
  secondary	
  dataset	
  
is	
  more	
  complicated.	
   	
   If	
  a	
  person	
  or	
  an	
  organization	
  provided	
  data,	
  should	
  they	
  
be	
   included	
   as	
   co-­‐author?	
   Based	
   on	
   substantial	
   contribution	
   definition	
   of	
   data	
  
authorship,	
   the	
   input	
   (secondary	
   dataset)	
   is	
   enough	
   for	
   consideration	
   for	
   the	
  
data	
   authorship.	
   Several	
   centers	
   have	
   policies	
   and	
   guidance	
   on	
   authorship	
  
although	
   some	
   of	
   the	
   details	
   above	
   are	
   not	
   always	
   adequately	
   covered	
   in	
   the	
  
guidance.	
  
Institutional	
  culture	
  	
  
Under	
   the	
   previous	
   structure	
   of	
   CGIAR,	
   the	
   former	
   Science	
   Council	
   conducted	
  
annual	
  evaluations	
  of	
  two	
  performance	
  indicators	
  for	
  CGIAR	
  research:	
  outcomes	
  
and	
   ex-­‐post	
   impact.	
   Output	
   indicators	
   strongly	
   focused	
   on	
   a	
   quantitative	
  
publication	
  matrix	
   and	
  were	
  directly	
   linked	
   to	
   allocation	
  of	
   funding.	
  Until	
   now	
  
annual	
  performance	
  evaluation	
  of	
  scientists	
   focus	
  strongly	
  on	
  publication	
  rates	
  
and	
  the	
  Hirsch-­‐index2is	
  used	
  as	
  a	
  standard	
  indicator	
  of	
  scientific	
  performance	
  in	
  
recruitment	
  procedures.	
  
When	
  evaluating	
  research	
  a	
  clear	
  distinction	
  should	
  be	
  made	
  between	
  research	
  
‘quality’	
   (i.e.	
   the	
   relative	
  excellence	
  of	
  academic	
  outputs	
   intended	
   for	
  academic	
  
consumption,	
   e.g.	
   journal	
   papers	
   and	
   books)	
   and	
   research	
   ‘impact’	
   (i.e.	
   the	
  
benefits	
   that	
   research	
  outcomes	
  produce	
   for	
  wider	
   society).	
  Unfortunately	
   this	
  
division	
   is	
   often	
   confused,	
   a	
   prime	
   example	
   being	
   when	
   journal	
   citation	
  
(‘quality’)	
  metrics	
   are	
   incorrectly	
  presented	
   as	
  measures	
   of	
   ‘impact’	
   (Donovan,	
  
2011).	
   	
  Even	
  when	
   journal	
   citations	
  are	
  used	
  correctly	
  as	
  a	
  measure	
  of	
  quality	
  
CGIAR	
  Centers	
  need	
  to	
  be	
  critical	
  about	
  what	
  is	
  to	
  be	
  measured.	
  Publications	
  are	
  
	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  
2	
  The	
  h-­‐index	
   is	
   an	
   index	
   that	
   attempts	
   to	
  measure	
   both	
   the	
   productivity	
   and	
  
impact	
  of	
  the	
  published	
  work	
  of	
  a	
  scientist	
  or	
  scholar.	
  
PAGE	
  -­‐	
  19	
  -­‐	
  
usually	
   reviewed	
  based	
  on	
   their	
   content,	
   their	
  originality	
  and	
   the	
  way	
  analysis	
  
and	
   interpretation	
   of	
   the	
   data	
   or	
   information	
   is	
   presented.	
   Publications	
   are	
  
seldom	
  evaluated	
  based	
  on	
  the	
  technical	
  rigor	
  of	
  the	
  data	
  collection	
  procedures,	
  
the	
   completeness	
   of	
   the	
   data	
   and	
   its	
   description,	
   and	
   alignment	
   with	
   existing	
  
community	
   standards.	
  To	
   translate	
   conceptual	
   frameworks	
   into	
   empirical	
  
sampling	
   designs	
   takes	
   significant	
   research	
   experience.	
   Thus	
   producing	
   a	
   high	
  
value	
  data	
  set	
  that	
  forms	
  the	
  basis	
  of	
  a	
  high	
  quality	
  scientific	
  publication	
  requires	
  
a	
  high	
  level	
  of	
  scientific	
  sophistication,	
  whereas	
  writing	
  the	
  paper	
  itself	
  requires	
  
a	
   good	
   grasp	
   of	
   language,	
   some	
   understanding	
   of	
   the	
   science	
   you're	
   writing	
  
about,	
  and	
  an	
  ability	
   to	
  "translate"	
   technical	
   information	
   into	
  plain	
  English	
  and	
  
write	
   about	
   it	
   compellingly	
   (Costandi,	
   2013).	
   	
   An	
   institutional	
   culture	
   that	
  
simplifies	
  research	
  quality	
  to	
  counts	
  of	
  publications	
  and	
  number	
  of	
  technologies	
  
released	
  does	
  not	
  nurture	
  the	
  sharing	
  of	
  data,	
  but	
  cultivates	
  protectionism	
  were	
  
data	
  is	
  viewed	
  as	
  the	
  intellectual	
  property	
  of	
  individual	
  researchers.	
  	
  
Exclusion	
   of	
   data	
   preparation	
   and	
   publication	
   from	
   the	
   research	
   project	
  
lifecycle	
  
Work	
   at	
   the	
   centers	
   is	
   still	
   very	
   much	
   project	
   driven	
   despite	
   the	
   reform	
   that	
  
aimed	
   at	
   strengthened	
   and	
   coordinated	
   funding	
   mechanisms	
   linked	
   to	
   the	
  
System	
   agenda	
   and	
   priorities.	
   As	
   such	
   project	
   lifecycles	
   are	
   shaped	
   by	
   grant	
  
requirements,	
  which	
  do	
  not	
  usually	
  see	
  data	
  as	
  part	
  of	
  the	
  project	
  deliverables.	
  
While	
  projects	
  adequately	
  account	
  for	
  the	
  costs	
  of	
  conducting	
  research,	
  including	
  
the	
   collection	
   and	
   analysis	
   of	
   research	
   data,	
   they	
   seldom	
   include	
   preparing	
   of	
  
data	
  and	
  metadata	
  and	
  curation	
  as	
  part	
  of	
  the	
  costs	
  of	
  the	
  research	
  process	
  led	
  
alone	
   long-­‐term	
  costs	
   for	
  data	
  storage	
  and	
  preservation	
  beyond	
   the	
   immediate	
  
lifetime	
   of	
   a	
   project.	
   Projects	
   are	
   usually	
   considered	
   closed	
   when	
   all	
   grants	
  
requirements	
   have	
   been	
   met,	
   leaving	
   preparing	
   of	
   data	
   and	
   metadata	
   and	
  
curation	
   unaccounted	
   and	
   unbudgeted	
   for.	
   Metadata	
   collection	
   and	
   proper	
  
documentation,	
  especially	
  of	
  multi-­‐country	
  data	
  sets	
  is	
  difficult	
  to	
  outsource	
  and	
  
overstretched	
  scientists	
  are	
  often	
  unable	
  to	
  allocate	
  the	
  necessary	
  time.	
  	
  As	
  data	
  
collection,	
   if	
   not	
   specifically	
   stated	
   as	
   project	
   output,	
   is	
   often	
   considered	
   as	
   a	
  
means	
   to	
   an	
   end,	
   especially	
   in	
   projects	
   with	
   a	
   strong	
   development	
   focus,	
   the	
  
necessary	
   scientific	
   scrutiny	
   for	
   sampling	
   design	
   and	
   theoretical	
   framework	
   is	
  
often	
  missing,	
  limiting	
  the	
  value	
  and	
  re-­‐use	
  of	
  the	
  data.	
  A	
  clear	
  recognition	
  of	
  the	
  
value	
   of	
   data	
   in	
   developmental	
   work	
   and	
   a	
   clear	
   mandate	
   to	
   make	
   the	
   data	
  
available	
  will	
  ensure	
  that	
  projects	
  will	
  use	
  their	
  scientific	
  expertise	
  to	
  create	
  high	
  
value	
  data	
  sets.	
  	
  
Disconnect	
  between	
  libraries	
  and	
  data	
  management	
  units	
  
Research	
  libraries	
  are	
  not	
  only	
  places	
  to	
  keep	
  collections,	
  but	
  their	
  real	
  strength	
  
and	
   power	
   lies	
   in	
   organizing,	
   preserving,	
   and	
   making	
   knowledge	
   accessible.	
  
Unfortunately,	
  libraries	
  and	
  data	
  units	
  in	
  the	
  Centres	
  are	
  positioned	
  at	
  opposite	
  
ends	
  of	
   the	
  research	
   lifecycle:	
  data	
  management	
  units	
  have	
  been	
  established	
  to	
  
help	
   researchers	
   collect	
   and	
   process	
   their	
   data,	
   and	
   libraries	
   to	
   support	
   the	
  
publications	
   that	
   result	
   from	
   research	
  projects.	
   Libraries	
   also	
  help	
   arrange	
   the	
  
search	
  for	
  publications	
  as	
  the	
  basis	
  for	
  new	
  research.	
  	
  While	
  both	
  are	
  considered	
  
to	
   be	
   research	
   support	
   units,	
   libraries	
   are	
   often	
   anchored	
   with	
   the	
  
communication	
  units	
  and	
  data	
  management	
  with	
   the	
   research	
  units;	
   each	
  with	
  
its	
   own	
   directorates,	
   reporting	
   lines	
   and	
   working	
   culture.	
   With	
   increasing	
  
numbers	
   of	
   journals	
   requesting	
   authors	
   to	
  make	
   the	
   replication	
   data	
   available	
  
PAGE	
  -­‐	
  20	
  -­‐	
  
and	
   the	
   rise	
   of	
   data	
   papers	
   the	
   functional	
   boundaries	
   of	
   librarian	
   and	
   data	
  
managers	
   is	
   becoming	
   blurred.	
  With	
   the	
   convergence	
   of	
   and	
   interdependency	
  
between	
  both	
  data	
  and	
  publications,	
  the	
  distinctive,	
  but	
  complimentary	
  skill	
  sets	
  
of	
   both	
   units	
   are	
   needed	
   to	
   safeguard	
   data	
   availability,	
   discoverability,	
  
interpretability	
  and	
  re-­‐useability.	
  	
  	
  	
  
	
  
Relative	
  benefits,	
  as	
  opposed	
  to	
  costs,	
  of	
  data	
  publishing	
  not	
  clear	
  
CGIAR	
  has	
  the	
  mandate	
  to	
  conduct	
  and	
  facilitate	
  research	
  in	
  development	
  with	
  a	
  
clear	
   target	
   to	
   reduce	
   poverty	
   and	
   hunger.	
   While	
   there	
   are	
   many	
   institutions	
  
with	
  a	
  similar	
  mandate,	
  very	
  few	
  have	
  a	
  similar	
  global	
  outreach	
  and	
  the	
  capacity	
  
to	
   collect	
   information	
   and	
  data	
   on	
   a	
   variety	
   of	
   food	
  production	
   systems	
   at	
   the	
  
household	
  or	
  farm	
  level.	
  While	
  it	
  seems	
  intuitively	
  that	
  making	
  these	
  information	
  
and	
   data	
   available	
   would	
   lead	
   to	
   more	
   rapid	
   advances	
   in	
   developmental	
  
questions,	
   it	
   is	
   less	
   clear	
   what	
   the	
   impact	
   pathways	
   of	
   data	
   sets	
   would	
   be.	
  	
  
Publishing	
  research	
  data	
  comes	
  with	
  an	
  increase	
  cost,	
  both	
  at	
  the	
  project	
  level	
  as	
  
well	
   as	
   at	
   the	
   center	
  management	
   level.	
   	
   Both	
   centers	
   and	
   donors	
   need	
   to	
   be	
  
assured	
  of	
  the	
  benefits	
  of	
  money	
  spend	
  on	
  data	
  publishing.	
   	
  Rather	
  than	
  simply	
  
assuming	
   that	
  open	
  access	
   is	
  beneficial	
   to	
   the	
  work	
  of	
  CGIAR	
  a	
   clear	
   theory	
  of	
  
change	
  with	
  an	
  appropriate	
  indicator	
  matrix	
  is	
  needed	
  to	
  allow	
  for	
  a	
  cost	
  benefit	
  
analysis	
   on	
   the	
   contribution	
   of	
   data	
   sets	
   to	
   CGIAR’s	
   overall	
   impact	
   along	
   the	
  
research	
  in	
  development	
  continuum.	
  	
  	
  
Recommendations	
  
Clear	
  mandate	
  to	
  include	
  data	
  as	
  research	
  outputs	
  
The	
  mandate	
  to	
  publish	
  research	
  data	
  as	
  research	
  outputs	
  needs	
  to	
  be	
  explicitly	
  
stated	
  in	
  the	
  Strategy	
  and	
  Results	
  Framework	
  (SRF),	
  which	
  sets	
  common	
  goals,	
  
strategic	
  objectives	
  and	
  results	
  to	
  be	
  jointly	
  achieved	
  by	
  CGIAR	
  and	
  its	
  partners.	
  
Centers	
  and	
  donors	
  have	
  to	
  recognize	
  data	
  sets	
  as	
  important	
  information	
  assets	
  
and	
  project	
  deliverables.	
  Only	
   if	
  data	
   sets	
  are	
   included	
   in	
   the	
  performance	
  and	
  
impact	
   evaluation	
   of	
   CGIAR,	
   will	
   the	
   centers	
   be	
   in	
   a	
   position	
   to	
   provide	
  
incentives	
  for	
  scientists	
  to	
  allocate	
  sufficient	
  time	
  to	
  data	
  quality	
  at	
  all	
  stages	
  of	
  
the	
   research	
   cycle;	
   from	
   development	
   of	
   conceptual	
   frameworks	
   consisting	
   of	
  
problem	
   definition,	
   hypothesis,	
   models	
   and	
   research	
   questions,	
   well-­‐
documented	
   research	
   designs	
   and	
  methods,	
   data	
  management	
   and	
   curation	
   to	
  
statistical	
  analysis.	
  	
  
CGIAR	
  Fund	
  allocation	
  to	
  support	
  cross-­‐center	
  collaboration	
  
While	
   research	
   data	
   are	
   produced	
   at	
   centre	
   level	
   and	
   shaped	
   by	
   the	
   research	
  
emphasis	
   and	
   character	
   of	
   each	
   centre	
   there	
   are	
   overarching	
   data	
   publishing	
  
issues	
   relevant	
   to	
   all	
   research	
   data	
   and	
   all	
   Centers.	
   Specific	
   CGIAR	
   Fund	
  
allocations	
  need	
  to	
  be	
  made	
  available	
  to	
  facilitate	
  cross-­‐center	
  collaboration	
  and	
  
information	
  sharing	
  with	
  respect	
  to	
  data	
  management	
  and	
  publishing.	
  Research	
  
activities	
   in	
   CGIAR	
   are	
   funded	
   through	
   the	
   CGIAR	
   Fund,	
   a	
   new	
   multi-­‐donor,	
  
multi-­‐year	
   funding	
   mechanism.	
   It	
   finances	
   research	
   aligned	
   with	
   the	
   Strategy	
  
and	
  Results	
   Framework	
  developed	
  by	
   the	
   CGIAR	
  Consortium	
  and	
   endorsed	
   by	
  
the	
   Funders	
   Forum	
   to	
   establish	
   common	
   goals,	
   objectives	
   and	
   results	
   for	
   the	
  
PAGE	
  -­‐	
  21	
  -­‐	
  
CGIAR	
  partnership.	
  The	
  Fund	
  is	
  already	
  facilitating	
  cross-­‐center	
  collaboration	
  in	
  
addressing	
   key	
   research	
   questions	
   (i.e.	
   gender,	
   capacity	
   development	
   and	
  
communications)	
   via	
   the	
   strategic	
   research	
   programs,	
   but	
   cross	
   cutting	
   issues	
  
like	
  data	
  sharing	
  do	
  not	
  have	
  a	
  home	
  yet.	
  	
  
Data	
  Policies	
  at	
  Centre	
  level	
  
Each	
  Center	
  should	
  have	
  a	
  research	
  data	
  management	
  policy	
  that	
  is	
  aligned	
  with	
  
the	
  CGIAR	
  Principles	
  on	
  the	
  Management	
  of	
  Intellectual	
  Assets,	
  the	
  CGIAR	
  Open-­‐
Access	
   Policy	
   and	
   consistent	
   with	
   the	
   aims	
   of	
   public	
   funded	
   research.	
   The	
  
Policies	
   need	
   to	
   lay	
   out	
   the	
   basic	
   principles	
   of	
   research	
   data	
   management	
   at	
  
center	
   level,	
  address	
  what	
  data	
  can	
  be	
  made	
  available	
   to	
   the	
  public	
  when,	
  how	
  
and	
   by	
   whom.	
   Roles	
   and	
   responsibilities	
   of	
   individual	
   scientists,	
   projects	
   and	
  
research	
   support	
   units	
   towards	
   this	
   afford	
   need	
   to	
   be	
   stated.	
   Legal	
   clarity	
   on	
  
licenses,	
   data	
   ownership	
   and	
   authorships	
   should	
  be	
   addressed.	
   Policies	
   should	
  
be	
   accompanied	
   by	
   flexible	
   implementation	
   guidelines	
   that	
   ensure	
   that	
   data	
   is	
  
used	
  by	
  the	
  Center	
  and	
  its	
  partners	
  in	
  the	
  most	
  efficient	
  way	
  while	
  safeguarding	
  
that	
  scientists	
  are	
  given	
  sufficient	
  time	
  to	
  produce	
  the	
  scientific	
  publications	
  they	
  
set	
  out	
  to	
  do.	
  	
  
Ethical	
  committee	
  to	
  be	
  established	
  in	
  all	
  Centers	
  
Each	
  Center	
  should	
  have	
  an	
  ethical	
  committee	
  that	
  is	
  responsible	
  for	
  developing	
  
the	
  centers	
  research	
  ethic	
  guidelines	
  and	
  policies	
  addressing	
  sensitive	
  data	
  and	
  
data	
   confidentiality,	
   as	
  well	
   as	
  appropriate	
  handling	
  of	
   research	
  data.	
   	
  Projects	
  
should	
  be	
  reviewed	
  based	
  on	
  their	
  adherence	
  to	
  the	
  accepted	
  ethical	
  standards	
  
of	
  a	
  genuine	
  research	
  study.	
  	
  
Clear	
  guidelines	
  on	
  authorship	
  attribution	
  
To	
   ensure	
   that	
   both	
   technical	
   as	
   well	
   as	
   scientific	
   staff	
   is	
   given	
   the	
   deserved	
  
credit	
   for	
   their	
   work	
   all	
   people	
   that	
   have	
   substantially	
   contributed	
   to	
   the	
  
creation	
   of	
   the	
   datasets	
   should	
   be	
   data	
   authors.	
   This	
   includes	
   all	
   people	
   that	
  
played	
  a	
  key	
  role	
  in	
  the	
  following:	
  	
  
1) Conceiving	
   and	
  designing	
   the	
   field	
  work	
   in	
   response	
   to	
   questions	
   of	
  
recognized	
  scientific	
   importance	
  and/or	
  relevance	
  for	
  developmental	
  
impact	
  and	
  policy	
  change.	
  
2) Development	
   and	
   implementation	
   of	
   research	
   designs,	
   choice	
   of	
  
methods,	
  quality	
  control	
  on	
  data	
  collection.	
  
3) Database	
  design,	
  data	
  cleaning,	
  validation	
  and	
  verification	
  processes	
  	
  
Zero	
  tolerance	
  of	
  scientific	
  fraud	
  
Each	
   Centre	
   needs	
   to	
   implement	
   zero	
   tolerance	
   policies	
   with	
   respect	
   to	
   data	
  
manipulation	
   and	
   should	
   have	
   explicit	
   standards	
   regarding	
   the	
   appropriate	
  
handling	
  of	
  research	
  data.	
  	
  
Adoption	
  of	
  OAI-­‐compliant	
  data	
  repositories	
  across	
  CGIAR	
  
The	
  Open	
  Archives	
   Initiative	
   Protocol	
   for	
  Metadata	
  Harvesting	
   (OAI-­‐PMH)	
   is	
   a	
  
low-­‐barrier	
  mechanism	
   for	
   repository	
   interoperability.	
   To	
   enable	
   the	
   broadest	
  
level	
  of	
  interoperability,	
  OAI-­‐PMH	
  mandates	
  that	
  metadata	
  should	
  be	
  exposed	
  as	
  
Dublin	
  Core.	
  Data	
  Providers	
  are	
  repositories	
  that	
  expose	
  structured	
  metadata	
  via	
  
PAGE	
  -­‐	
  22	
  -­‐	
  
OAI-­‐PMH.	
  Service	
  Providers	
  then	
  make	
  OAI-­‐PMH	
  service	
  requests	
  to	
  harvest	
  that	
  
metadata.	
  By	
  using	
  OAI-­‐PMH	
  compliant	
  repositories	
  centers	
  can	
  have	
  their	
  own	
  
individual	
   institutional	
   repositories	
   each	
   with	
   their	
   own	
   particular	
   collection	
  
policies	
   and	
   administrative	
   systems,	
   but	
   to	
   be	
   linked	
   into	
   one	
   large,	
   a	
   virtual,	
  
global	
  repository	
  through	
  the	
  use	
  of	
  the	
  OAI-­‐PMH.	
  	
  
Linking	
  data	
  and	
  publications	
  
To	
   improve	
   scientific	
   publications,	
   consensus	
   with	
   scientific	
   peers	
   and	
   public	
  
trust	
   in	
   the	
   quality	
   of	
   our	
   research	
   outputs	
   each	
   Centre	
   should	
   make	
   all	
  
necessary	
  raw	
  data	
  public	
  to	
  reproduce	
  or	
  replicate	
  every	
  scientific	
  publication	
  
that	
  is	
  based	
  on	
  research	
  data.	
  	
  
Building	
  libraries	
  capacity	
  for	
  data	
  curation	
  	
  
The	
  Centers	
  libraries	
  need	
  to	
  embrace	
  the	
  challenges	
  that	
  come	
  with	
  publishing	
  
research	
   data.	
   Libraries	
   need	
   to	
   support	
   data	
   units	
   in	
   their	
   efforts	
   to	
   publish	
  
research	
   data	
   by	
   providing	
   persistent	
   identification/citation	
   of	
   datasets,	
  
guidance	
  on	
  authorships	
  and	
  solutions	
  for	
  data	
  description,	
  documentation	
  and	
  
retrieval,	
   which	
   together	
   facilitate	
   findability	
   (Reilly,	
   2012).	
   They	
   must	
   also	
  
ensure	
   long-­‐term	
  data	
  archiving,	
   including	
  data	
   curation	
  and	
  preservation	
  as	
  a	
  
condition	
  for	
  data	
  interpretability	
  and	
  re-­‐usability.	
  The	
  Centers	
  need	
  to	
  invest	
  in	
  
developing	
   the	
   staff	
   skills	
   required	
   for	
   achieving	
   the	
   data	
   curation	
   role	
   in	
   the	
  
libraries	
   and	
   need	
   to	
   of	
   recruit	
   library	
   staff	
   with	
   experience	
   in	
   research	
  
disciplines.	
  
Specific	
  funds	
  to	
  publish	
  legacy	
  data	
  
Legacy	
   data	
   sets	
   that	
   have	
   high	
   potential	
   to	
   contribute	
   towards	
   achieving	
   the	
  
four	
  system	
  level	
  outcomes	
  should	
  be	
  archived	
  and	
  published.	
  Most	
  of	
  these	
  data	
  
sets	
  are	
  fully	
  documented,	
  however	
  documentation	
  and	
  data	
  formats	
  need	
  to	
  be	
  
brought	
   in	
   line	
  with	
   today’s	
   requirements	
   and	
   standards.	
  To	
  ensure	
   that	
   these	
  
data	
  sets	
  can	
  be	
  made	
  available	
  to	
  CGIAR	
  scientists	
  and	
  partners,	
  working	
  in	
  the	
  
CGIAR	
  Research	
  Programs,	
  specific	
  financial	
  incentives	
  need	
  to	
  be	
  provided.	
  	
  	
  	
  
Changing	
  institutional	
  culture	
  
Performance	
  evaluations	
  both	
  at	
  individual	
  scientist	
  level	
  as	
  well	
  as	
  center	
  level	
  
need	
  to	
  shift	
  from	
  using	
  simplistic	
  indicators	
  metrics	
  such	
  as	
  numbers	
  of	
  papers,	
  
positions	
   in	
   lists	
  of	
  authors,	
  and	
   journals’	
   impact	
   factors	
   towards	
  assessing	
   the	
  
quality	
  of	
  research	
  itself.	
  Centers	
  and	
  science	
  managers	
  need	
  to	
  put	
  performance	
  
indicators	
   in	
   place	
   that	
   not	
   only	
   reward	
   the	
   excellent	
   scientific	
   writers	
   the	
  
system	
   has,	
   but	
   also	
   the	
   scientific	
   and	
   technical	
   excellence	
   that	
   leads	
   to	
   the	
  
creation	
  of	
  the	
  data,	
  methods	
  and	
  ideas	
  that	
  are	
  supposed	
  to	
  be	
  communicated	
  in	
  
the	
  papers.	
  Internal	
  project	
  reviews	
  should	
  take	
  into	
  account	
  the	
  technical	
  rigor	
  
of	
   the	
   data	
   collection	
   procedures,	
   the	
   completeness	
   of	
   the	
   data	
   and	
   its	
  
description,	
   and	
   alignment	
   with	
   existing	
   community	
   standards.	
  	
   Scientists	
   and	
  
their	
  field	
  teams	
  should	
  be	
  encouraged	
  to	
  produce	
  peer-­‐reviewed	
  data	
  papers.	
  
	
  
	
  
	
   	
  
PAGE	
  -­‐	
  23	
  -­‐	
  
Key	
  references	
  
AfricaRice	
  Dataverse	
  	
  http://africarice.org/warda/dataverse.asp	
  	
  
Alsheikh-­‐Ali	
  AA,	
  Qureshi	
  W,	
  Al-­‐Mallah	
  MH,	
  Ioannidis	
  JPA	
  (2011)	
  Public	
  
Availability	
  of	
  Published	
  Research	
  Data	
  in	
  High-­‐Impact	
  Journals.	
  PLoS	
  ONE	
  
6(9):	
  e24357.	
  doi:10.1371/journal.pone.0024357	
  	
  
Altman	
  DG,	
  Furberg	
  CG,	
  Grimshaw	
  GM,	
  Rothwell	
  PM	
  (2006).	
  Lead	
  editorial:	
  
Trials	
  –	
  using	
  the	
  opportunities	
  of	
  electronic	
  publishing	
  to	
  improve	
  the	
  
reporting	
  of	
  randomized	
  trials.	
  Trials,	
  7:6.	
  doi:	
  10.1186/1745-­‐6215-­‐7-­‐6.	
  
[PMC	
  free	
  article]	
  [PubMed]	
  [Cross	
  Ref]	
  
Anon	
  (2008)	
  Improving	
  Research	
  Data	
  Management	
  and	
  Sharing	
  in	
  the	
  Alliance	
  
of	
  CGIAR	
  Centres;	
  A	
  Working	
  Paper	
  for	
  Consideration	
  of	
  the	
  CGIAR	
  Alliance,	
  
online	
  
http://cropwiki.irri.org/gcp/images/4/40/ResearchDataManagement_CG-­‐
ADE.pdf	
  	
  
Bioversity	
  International	
  Dataverse	
  (2013),	
  
http://thedata.harvard.edu/dvn/dv/Bioversity	
  
CIAT	
  Dataverse,	
  http://dvn.iq.harvard.edu/dvn/dv/CIAT	
  
Costandi	
  M	
  (2013)	
  A	
  good	
  story	
  conveys	
  wonderment,	
  The	
  Guardian,	
  Monday	
  22	
  
April	
  2013,	
  online	
  	
  http://www.theguardian.com/science/2013/apr/22/mo-­‐
costandi-­‐science-­‐writing	
  
Donovan	
  C	
  (2011)	
  Impact	
  is	
  a	
  strong	
  weapon	
  for	
  making	
  an	
  evidence-­‐based	
  case	
  
for	
  enhanced	
  research	
  support	
  but	
  a	
  state-­‐of-­‐the-­‐art	
  approach	
  to	
  
measurement	
  is	
  needed.	
  Citations,	
  REF	
  2014,	
  Research	
  funding,	
  online	
  
http://blogs.lse.ac.uk/impactofsocialsciences/2011/08/22/impact-­‐strong-­‐
weaponevidence-­‐based-­‐case-­‐for-­‐enhanced-­‐research-­‐support-­‐but-­‐a-­‐state-­‐of-­‐
the-­‐art-­‐approach-­‐to-­‐measurement-­‐is-­‐needed	
  
FAO/Bioversity	
  International	
  (2012)	
  FAO/Bioversity	
  Multi-­‐Crop	
  Passport	
  
Descriptors	
  V.2	
  [MCPD	
  V.2].	
  11p.	
  	
  CGIAR	
  Principles	
  on	
  the	
  Management	
  of	
  
Intellectual	
  Assets	
  (2012)	
  
http://www.cgiarfund.org/sites/cgiarfund.org/files/Documents/PDF/cgiar_
principles_management_intellectual_assets_7march_2012.pdf	
  	
  	
  
Graf	
  C,	
  Wager	
  E,	
  Bowman	
  A,	
  Fiack	
  S,	
  Scott-­‐Lichter	
  D,	
  Robinson	
  A	
  (2007)	
  Best	
  
practice	
  guidelines	
  on	
  publication	
  ethics:	
  a	
  publisher's	
  perspective.	
  
International	
  journal	
  of	
  clinical	
  practice	
  61(152):	
  1-­‐26.	
  
Hartter	
  J,	
  Ryan	
  SJ,	
  MacKenzie	
  CA,	
  Parker	
  JN,	
  Strasser	
  CA	
  (2013)	
  Spatially	
  Explicit	
  
Data:	
  Stewardship	
  and	
  Ethical	
  Challenges	
  in	
  Science.	
  PLoS	
  Biol	
  11(9):	
  
e1001634.	
  doi:10.1371/journal.pbio.1001634	
  	
  
Hijmans	
  RJ,	
  Cameron	
  SE,	
  Parra	
  JL,	
  Jones	
  PG,	
  Jarvis	
  A	
  (2005)	
  Very	
  high	
  resolution	
  
interpolated	
  climate	
  surfaces	
  for	
  global	
  land	
  areas.	
  International	
  journal	
  of	
  
climatology	
  25(15):	
  1965-­‐1978.	
  
Hoddinott	
  J,	
  Yohannes	
  Y	
  (2011)	
  Ethiopian	
  Rural	
  Household	
  Surveys	
  (ERHS),	
  
http://hdl.handle.net/1902.1/15646	
  UNF:5:k2eYxsY6t/jVXblm/UAkRg==	
  
International	
  Food	
  Policy	
  Research	
  Institute	
  [Distributor]	
  V7	
  [Version]	
  
ICRISAT Dataverse http://dataverse.icrisat.org/dvn 
Reilly, S (2012) The Role of Libraries in Supporting Data Exchange 
http://conference.ifla.org/past/2012/116-reilly-en.pdf. 
PAGE	
  -­‐	
  24	
  -­‐	
  
IFPRI	
  Dataverse	
  http://thedata.harvard.edu/dvn/dv/IFPRI	
  
IMWI Water Data Portal 	
  http://waterdata.iwmi.org/ 
International	
  Water	
  Management	
  Institute	
  (2011)	
  Research	
  Data	
  Management	
  
Policy	
  And	
  Implementation	
  Guideline	
  
Muraya	
  P,	
  Coe	
  R	
  (2001)	
  Research	
  Discussion	
  Paper	
  2:	
  Looking	
  after	
  our	
  
investments:	
  Improving	
  Research	
  Data	
  Management	
  in	
  ICRAF,	
  
https://sites.google.com/a/cggmail.org/cgiar-­‐data-­‐management-­‐
meeting/resource-­‐documents	
  
ODE	
  Report	
  on	
  Integration	
  of	
  Data	
  and	
  Publications	
  (2011)	
  
http://www.alliancepermanentaccess.org/wp-­‐
content/uploads/downloads/2011/11/ODE-­‐
ReportOnIntegrationOfDataAndPublications-­‐1_1.pdf	
  
Quisumbing	
  A,	
  Baulch	
  B	
  (2010)	
  Chronic	
  Poverty	
  and	
  Long	
  Term	
  Impact	
  Study	
  in	
  
Bangladesh	
  http://hdl.handle.net/1902.1/17045	
  	
  
UNF:5:8MUn92HhwQhRKF69wSTwaA==	
  International	
  Food	
  Policy	
  Research	
  
Institute	
  [Distributor]	
  V5	
  [Version]	
  
Tenopir	
  C,	
  Allard	
  S,	
  Douglass	
  K,	
  Aydinoglu	
  AU,	
  Wu	
  L,	
  Wu	
  L,	
  Read	
  E,	
  Manoff	
  M,	
  and	
  
Frame	
  M	
  (2011)	
  Data	
  Sharing	
  by	
  Scientists:	
  Practices	
  and	
  Perceptions.	
  PLoS	
  
ONE	
  6(6):	
  e21101.	
  doi:10.1371/journal.pone.0021101	
  	
  
World	
  Agroforestry	
  Centre	
  -­‐	
  ICRAF	
  Dataverse	
  (2011),	
  
http://thedata.harvard.edu/dvn/dv/icraf	
  
World	
  Agroforestry	
  Centre	
  -­‐	
  ICRAF	
  Geoportal	
  (2012),	
  
http://geoportal.worldagroforestry.org/	
  
	
  
PAGE	
  -­‐	
  25	
  -­‐