WEBVTT Kind: captions Language: en 00:00:01.120 --> 00:00:03.120 Thanks so much for inviting me.   00:00:03.120 --> 00:00:07.440 I want to start by recognizing that this  was a team effort and we had a great team.  00:00:12.320 --> 00:00:16.720 Crises like the pandemic emerge and we  find ourselves in urgent need of data.   00:00:16.720 --> 00:00:22.560 Sometimes the data we need are about biodiversity.  In this case, we'd like to know things about bats.   00:00:23.600 --> 00:00:28.960 In other cases, like oil spills, it might  be the entire biota in a particular region   00:00:28.960 --> 00:00:34.960 about which we need data. Until our work,  we didn't have a set of crisis response   00:00:34.960 --> 00:00:43.840 protocols to rapidly enhance data about an  important source of biodiversity information:   00:00:46.240 --> 00:00:49.840 that is the world's 3 to 4  billion biodiversity specimens.  00:00:50.720 --> 00:00:55.120 As we see in our current pandemic, it could be  that a narrow subset of those specimens suddenly   00:00:55.120 --> 00:01:03.360 become critical to crisis response. Specimens have  associated information that documents the what was   00:01:03.360 --> 00:01:09.680 collected, where it was collected, who collected  it, and other information. There are also time   00:01:09.680 --> 00:01:15.840 capsules of potential information, since genomic  data can often be derived from the specimen or   00:01:15.840 --> 00:01:22.240 its disease-causing agents. These are just a  few specimens of the species of Horseshoe bat   00:01:22.240 --> 00:01:28.560 in which the closest relative of SARS-CoV-2  has been found. That is: Rhinolophus affinis.  00:01:31.760 --> 00:01:35.280 We targeted a set of three  closely-related families,   00:01:35.280 --> 00:01:40.960 including the family of Rhinolophus  affinis for specimen data enhancement.  00:01:44.880 --> 00:01:50.640 These are the mappings of Horseshoe bat specimens  at the two major aggregators of specimen data.   00:01:51.360 --> 00:01:56.320 I want to emphasize that data coming from  collections and served by these aggregators   00:01:56.320 --> 00:02:01.120 are valuable in their current state. However,  the data have some qualities that can be   00:02:01.120 --> 00:02:06.960 improved through consideration of the data in  aggregate and the data have been created over   00:02:06.960 --> 00:02:12.000 two or more decades, meaning that the data  have not all benefited from our current   00:02:12.000 --> 00:02:16.880 understanding of best practices and the  availability of software to improve some steps.  00:02:20.480 --> 00:02:25.440 We focused on enhancing the data in these ways  and I'll walk through those in bold with you.   00:02:26.080 --> 00:02:31.520 If you pay attention to the change in slide title  you can follow our relatively rapid progression   00:02:31.520 --> 00:02:37.200 through these activities. The specimen data coming   00:02:37.200 --> 00:02:41.200 from the two major aggregators has  overlapped but it's not identical.   00:02:42.000 --> 00:02:47.200 De-duplicating their records produced  about 90,000 in scope records.  00:02:49.680 --> 00:02:56.960 The records are curated by 118 institutions  worldwide. The top 10 of these institutions   00:02:56.960 --> 00:03:06.640 together share 63% of the records. We could only assign or assess coordinates   00:03:06.640 --> 00:03:12.560 for collections - collection locations when  those locations are described in the shared data.   00:03:13.120 --> 00:03:16.400 And about two-thirds of the  records had that information.   00:03:17.440 --> 00:03:22.560 Of those, about two-thirds arrived with  pre-assigned coordinates and one third did not.   00:03:25.680 --> 00:03:28.640 We were able to assess or  assign coordinates in 95%   00:03:29.200 --> 00:03:34.800 of the total possible cases and we modified  pre-existing coordinates about half the time.   00:03:36.160 --> 00:03:40.560 The median amount that a pre-existing  coordinate was moved was six kilometers.  00:03:44.640 --> 00:03:52.560 Importantly, the relevant metadata fields went  from mostly empty to mostly complete with such   00:03:52.560 --> 00:03:59.840 useful information added as geo-referencing  protocol and geo-referencing resources.  00:04:01.680 --> 00:04:04.720 In this summary, at the country  level, you can see where the   00:04:04.720 --> 00:04:08.800 greatest number of specimens have been  collected by the size of the pie chart   00:04:08.800 --> 00:04:13.840 and the relative number of new coordinates  added to the specimens from those countries.  00:04:17.680 --> 00:04:22.080 Here are the coordinates for collecting  locations for each of our focal families.  00:04:26.480 --> 00:04:31.840 We compared our coordinates with prior range  maps for the species when they were available   00:04:31.840 --> 00:04:36.960 from the International Union for Conservation  of Nature (IUCN). Here's an example range   00:04:36.960 --> 00:04:42.240 circumscription in red for one species,  this is again Rhinolophus affinis,   00:04:42.240 --> 00:04:48.400 and our coordinates for that species in green.  We found that georeferenced specimens suggest   00:04:48.400 --> 00:04:55.840 range extensions for 153 of the 169 focal bat  taxa for which we have these kinds of maps.   00:04:56.880 --> 00:05:02.160 This is a significant, significant expansion  of our understanding of where to find the bats.   00:05:03.920 --> 00:05:11.280 This is a screenshot of a web-based horseshoe  bat data explorer for IUCN map assessors   00:05:11.280 --> 00:05:16.880 and other stakeholders to look at locality  coordinates relative to the current IUCN maps,   00:05:18.320 --> 00:05:23.840 with links back to complete records in our system.  00:05:25.200 --> 00:05:31.680 The records arrived with 2,930 distinct values  referencing people who collected or identified   00:05:31.680 --> 00:05:38.160 the specimens. We were able to assign 803  unique identifiers to a subset of those values.   00:05:39.200 --> 00:05:43.120 These unique identifiers, or ORCID  IDs when the person is living,   00:05:43.120 --> 00:05:52.720 and Wikidata QIDs when the person is deceased.  An additional 437 values representing 359 people   00:05:52.720 --> 00:05:57.760 are reasonably assigned to persons currently  living but who do not yet have an ORCID ID.  00:06:00.800 --> 00:06:08.240 To do this we engage 34 people who are mostly  bat experts from 13 countries. These experts and   00:06:08.240 --> 00:06:13.600 our data curators found that they could associate  about half of the records to a unique identifier   00:06:13.600 --> 00:06:19.840 for specimen collector and about two-thirds  of those for specimen identifier.  00:06:23.840 --> 00:06:28.400 The value of doing this for a crisis response  might not be immediately apparent, but it could   00:06:28.400 --> 00:06:35.760 be among the most important things that we did. We  identified 117 living people with ORCID IDs with   00:06:35.760 --> 00:06:41.760 experience collecting the bats. You might say that  it's easy to find the bat experts - just approach   00:06:41.760 --> 00:06:47.360 their professional societies or do a literature  search. However, the bat collectors and those   00:06:47.360 --> 00:06:53.360 you might find in that way will only partially  overlap. The bat collectors turned out to include   00:06:53.360 --> 00:06:58.800 a considerable diversity of professions including  those who are not professional biologists.   00:06:59.840 --> 00:07:04.880 Here are a few other descriptors - descriptions  of collectors with valuable experience doing   00:07:04.880 --> 00:07:13.600 field work in sometimes remote areas. Remember,  excuse me, remember we also identified 359 living   00:07:13.600 --> 00:07:19.520 bat collectors who don't have ORCID IDs.  Together, this is a rolodex of potential   00:07:19.520 --> 00:07:24.880 contacts for those of you who need to go back  into the field to relocate bat populations.  00:07:27.920 --> 00:07:33.120 Prior to data enhancement 5.5% of the records  had information about associated sequences.   00:07:33.760 --> 00:07:39.360 We identified an additional about  1,100 specimens with which we could   00:07:39.360 --> 00:07:47.600 associate new sequences that we found. Our versioned data, and importantly our   00:07:48.560 --> 00:07:54.800 protocols, are shared at Zenodo so that we have  now blazed a trail that others can follow for   00:07:54.800 --> 00:08:01.360 rapid data enhancement of specimens during the  next crisis. We expect to share our final versions   00:08:01.360 --> 00:08:06.880 of everything there very soon. However, I will  note that the current version that's up right   00:08:06.880 --> 00:08:12.160 now is very close to the final version. We're  close to submission of a manuscript focused on   00:08:12.160 --> 00:08:17.520 the work and expect to make the Horseshoe bat data  explorer broadly available to those who need it.  00:08:19.120 --> 00:08:26.080 The EU recently announced new funding for  the creation of records, About 20 - about   00:08:27.280 --> 00:08:36.000 approximately 20,000 bat specimens, and we expect  the foundation that we laid to speed up that work.  00:08:39.200 --> 00:08:44.400 I want to thank those who contributed their  time and expertise to the people disambiguation   00:08:44.400 --> 00:08:49.920 and in the first paragraph, but not  assigned ORCID IDs there for lack of space.   00:08:50.800 --> 00:08:52.960 And thank you to the NSF for supporting the work.