Collecting quantitative metadata by counting all specimens in a herbarium

Transcript

1. Collec&ng   quan&ta&ve  metadata   by  coun&ng  all  specimens  in

    a   herbarium   Peter   Desmet  

2. Quan&ta&ve  metadata  are  cool!   A  very  colourful  presenta&on  by

     @peterdesmet  #tdwg  

3. Index  Herbariorum   350,000,000   herbarium  specimens  worldwide  

4. 25,000,000   digi&zed  and  published  (=  7%)   GBIF  Data

    Portal  (Andrea  Hahn)  

5. What  do  we  know  about  the  other   93%  ?

    

6. Descrip&ve   metadata  

7. Metadata  registries   bit.ly/IH-­‐herbaria   biocol.org  

8. Collec&on  name  +  code   Address   Staff   Subcollec&ons

    

9. Es&mated  size   Based  on  what?   Actually  counted?  

10. Geographic  scope   Pre^y  well  described   How  distributed?  

11. Taxonomic  scope   Vascular  plants  +  Bryophytes?   Families?  

    Genera?  

12. Can  we  get  some  real  numbers?  

13. Vascular  plants  specimens   are  organized  in   Folders  

14. None
15. None
16. None
17. None

18. What  if  we  counted  the  folders?  

19. And  the  #  of  specimens  per  folder?  

20. ?  $   How  much  would  it  cost?  

21. ?  days   How  long  would  it  take?  

22. What  we  did  at  the   Marie-­‐Victorin   Herbarium  (MT)

     

23. Move  an  es&mated  900,000  specimens  

24. More  space   Reassign  350  -­‐>  640  cases  

25. New  classifica&on   Flowering  plants:  APG  III  (2009)   Ferns:

     Smith  et  al.  (2006)  

26. Coun&ng   Digi&zing   Data  cleaning   Publishing  

27. Coun&ng  

28. None

29. Average  age  >  60  

30. None

31. 1  summer  

32. 826  work  hours   110  work  days,  22  work  weeks

     

33. Digi&zing  

34. 4  volunteers  

35. Paper  -­‐>  Excel  

36. Data  cleaning  

37. 2  volunteers   1  professor   1  informa&cian  

38. Correc&ng  errors   Typos,  missed  genera,  dubious  counts  

39. New  classifica&on   Assigning  families,  correc&ng  genera  

40. Format  data  

41. Publishing  

42. 1  informa&cian   (me)  

43. Google  Fusion  Tables   bit.ly/mt-­‐inventory-­‐gk  

44. Darwin  Core  Archive  via  IPT   bit.ly/mt-­‐inventory  

45. Metadata  =  EML   Descrip&ve  metadata  

46. Occurrence  dataset   basisOfRecord  =  PreservedSpecimen  

47. 1  record   1  folder   1  genus   1

     loca&on   in  1  tray  

48. #  specimens   individualCount  

49. What  do  we   know  now?  

50. 22,298   folders  

51. 628,664   specimens  

52. 2/3   of  previous  es&mate  

53. 21.5%   digi&zed  

54. 380   families  

55. 82%   of  known  families  

56. 5,298   genera  

57. 6   con&nents  

58. Combina&ons   Rubus  specimens  from  Canada?   Yes:  2921,  in

     trays  A236-­‐07  –  A238-­‐04    

59. Useful  for  us   In-­‐house  management  &  planning   Digi&za&on

     priori&es  

60. Useful  for  others?   Loans   Demand  driven  digi&za&on?  

61. Granularity   Genus,  con&nent  -­‐>  Useful  for  climate   change

     &  invasive  species  studies?  

62. Global  picture   Really  350  mil.  specimens?   How  distributed

     over  genus  &   con&nent?  

63. Cost  /  Time  ?  

64. 158   work  days   Publishing  1%   Data  cleaning

     21%   Digi&zing  8%   Coun&ng  70%  

65. 5,740  $   total  salary  cost   Publishing  7%  

    Digi&zing  0%   Coun&ng  37%   Data  cleaning  56%  

66. 110  specimens  =  1$   100  &mes  cheaper  than  full

     digi&za&on  

67. 3,200,000  $   All  350  mil.  specimens  

68. 138  h     1049  h   Staff  5,740  $

      Volunteers  0  $   88%  by  volunteers  

69. 16,230  $   10$  wage  for  “volunteers”  +  staff  salary

     

70. 9,000,000  $   All  350  mil.  specimens  

71. 340  years   1  person  at  7.5h/day,  5  days/week,  no

      holidays  

72. 26  days   One  person  per  herbarium   3,400  herbaria

     -­‐  Index  Herbariorum  

73. ?!   Tricky  to  extrapolate!   What  about  non-­‐mounted  specimens?

      How  useful  is  this  data?   Is  there  a  metadata  repository?  

74. First  step   Towards  some  real  numbers  

75. Thanks!   bit.ly/mt-­‐inventory   Peter   Desmet