Transcript Using OBIS to Provide Reliable Regional Scale Estimates of

Using OBIS to Provide Reliable
Estimates of Population Indices
for Marine Species from Research
Trawl Surveys
Ocean Biodiversity Informatics Conference
Hamburg Germany, November 29, 2004
Robert M. Branton1, Daniel Ricard2
1Bedford Institute of Oceanography, 2Dalhousie University
Nova Scotia, Canada
[email protected]
Background
• Traditionally trawl surveys are species rich
(100s) with analysis focus on commercial
species (~10s).
• Recently other species have been added
to sampling protocols thus enabling
investigation of ecosystem issues.
• OBIS is expected to provide a basis for
interoperability of these data with other
scientific disciplines.
Presentation focus:
1.
2.
3.
4.
DFO/NOAA trawl data
Trawl data quality
DFO’s OBIS provider service
Ways to improve OBIS
DFO - Canada Dept. of Fisheries and Oceans
NOAA - USA National Ocean and Atmospheric Agency
1) DFO/NOAA trawl data
• ECNASAP database
created in 1995 :
–
–
–
–
Year = 1988
5 laboratories
1970-94
50,000 fishing sets
276 species.
• Static copy of DFO
4VWX summer
survey posted on
OBIS in 2002.
ECNASAP – East Coast North America
Strategic Assessment Project
Maps produced using DFO - ACON
Status
• ECNASAP available from DFO/NOAA staff:
– 300+ column SPSS file on CD
– not updated since ‘95 although surveys are ongoing.
• DFO now testing OBIS views for ECNASAP and
various other DFO trawl surveys:
– basis for ‘Status of Ecosystem Reports’.
• OBIS-ECNASAP exported to U. of Southern Maine:
– basis for ‘Gulf of Maine Ocean Data Partnership’.
Survey
Metadata
Database
Content
Fishing
Fishing Set
Set
Metadata
Metadata
Area
Metadata
Observed
Observed
Numbers
Numbers and
and
Weights
Species
Metadata
DFO/NOAA
ECNASAP
OBIS
Specimen
Details
Length
Counts
Sex
And Maturity
Age
Readings
Parasites and/or
Stomach Content
2) Trawl Data quality
• Validate survey species lists using ITIS:
– Get most current scientific names and hierarchies
– Taxonomists / survey staff review discrepancies,
correct lists, note species difficult to identify and/or not
routinely sampled.
• Add hierarchy data to species lists:
– Use cumulative discovery curves for each hierarchy
level to investigate protocol changes vs. new
discoveries in survey area.
ITIS - Integrated Taxonomic Information System
Taking care of zeros
• Zero catches & counts
usually not recorded:
– not looking = NULL,
– looking but not finding = 0.
• Zeros are important when
mapping distribution and
calculating averages.
• Augment species list to
indicate when and where
to assume zero.
0.255 / tow
Average = 2.405
Standardizing Observations
• General recommendations:
– Define Collection Codes for each survey vessel, sampling
gear, stratification plan & season combination (series)
– Adjust Observed Individual Count and Weight (at length,
sex & maturity) by sampling ratio (e.g. total/sample)
– Don’t include damaged sets.
• Indicate if data are standardized or estimated:
– fishing sets for distance towed (e.g. standard/observed),
species for catchability by gear (e.g. proportion caught at
length), …
– numbers at age using stock specific age length keys (e.g.
proportion at age for given length), …
Confidence Limits
– give links to citable
publications.
50
Aerage Weight per Standard Tow (Kgs)
• Relative indices such as
‘average per standard
tow’ should include
variance or standard
error.
• Absolute estimates such
as ‘total biomass’ and
‘total abundance’ should
also be peer reviewed.
40
30
20
10
0
1990
1992
1994
1996
1998
Year
2000
2002
2004
3) DFO’s OBIS provider service
• Inputs
– FTP for small databases (e.g. museums)
– SQL*net for large databases (e.g. research labs)
• Output
– DiGIR XML to OBIS global cache portal at
Rutgers U.
Architecture Diagram
Small Remote
Data Bases
FTP
BIO DMZ
ODBC
OBIS
Portal
DiGIR
Provider
SQL*net
Large Local
Data Bases
MSAccess
DBs
DMZ
Oracle
DB
SQL*net
Large Remote
Data Bases
DFO Firewall
Remote
Oracle DBs
Pre-Scheduled
Exports Only
Operational
Oracle
DBs
Trawl Survey & Ecosystem Reporting
http://www.dfo-mpo.gc.ca/CSAS/CSAS/English/Research_Years/2003/2003_089_E.htm
4) Ways to improve OBIS
Enhance existing schema concepts:
• Basis of Record – stratum average, stock estimate
• Locality – stratum, ecozone, grid square, stock area
• Life Stage – maturity stage, age class.
Add new concepts to schema:
•
•
•
•
•
Number of Samples and Sampling Units in Locality
Length Class of Observed Individuals
Observed Individual Count Variances or Error Estimate
Observed Weight Variances or Error Estimate
Parent Catalog Number (for stomach contents and parasites).
…/
Add new schemas:
• Collection/survey metadata
• Area gazetteer (e.g. stratum, ecozone, grid square and
stock areas).
Enhance end-user interface:
• Collection based multi-species mapping and reporting
– expanding pie symbol maps (shown earlier)
– species by row or column
• missing values as zeroes or nulls
• species catchability standardization
• summary statistics by stratum, ecozones, ...
Expected Benefits
• Systematic basis for ongoing
enhancement and extension of the OBIS
schema and interface.
– Ability to integrate data from disparate
sampling schemes
– Capacity to derive population/community
indices of abundance, diversity, production,
etc. around the world.
e.g. Trophic Cascade models (using Trawl
and CPR data) being developed for Scotian
Shelf could also be tested on North Sea.
Compliments of
Ken Frank, DFO