Described are the methods used to identify and enumerate seabirds in a vessel-based census, as well as basic analytical methods used to date to summarize and examine seabird relationships with oceanographic and biological factors.
Visual observations of seabird distribution and abundance using strip-transect methodology.
2. Seabird Surveys
One trained observer identified and counted birds from the flying bridge or pilot house of the ship during all daylight hours, excluding periods of inclement weather. Surveys were conducted while the vessel was underway at >5 knots and sustained winds at <25 knots. Observations were made with binoculars (7x – 10x) or with the naked eye.
A hand-held range finder (Heinemann 1981) was used to ground-truth the width of the survey strip.
Seabirds that entered a 90-degree arc from the bow to the beam of the ship, and out to 300 m on the one side of the ship with the best visibility (i.e., least sun glare) were identified to lowest taxonomic level possible, enumerated, and assigned to one of five behavioral codes (flying, sitting on the water, feeding, ship-following, or milling). Data are logged into a field computer using the program DLog (R.G. Ford Consulting) and older data recorded using program FLK (G.L. Hunt, pers. comm.).
Ship-following individuals were recorded when first sighted and then ignored thereafter (e.g., albatrosses).
The position of the ship was automatically time-coded into the survey log every 10 seconds using the ship’s GPS system.
3. Data Summary
Survey data were grouped into “bins”, generally 3km in length; the centroid of each bin was geo-referenced.
Seabird abundance data were collapsed and filtered for the behaviors “in flight includes ship following” and “sitting on the water (includes feeding and milling)” by species and date.
Abundance data were summed across each survey day to achieve a daily total; the daily total was divided by the area surveyed (km2) to produce a density estimate for each day. Note: ‘day’ was selected as the basic sampling unit to avoid problems of spatial autocorrelation in assessing seabird density per survey. Certainly, other summarization procedures are possible and should be used to meet specific project objectives.
An average daily density of birds was calculated (number birds/km2) per species per behavior (noted above).
Overall species-specific relative abundance is expressed as the number of individuals sighted per unit area surveyed (birds/km2) over the entire survey period by averaging daily density estimates.
4. Spatial Data Analysis
The distribution and abundance of seabirds is related to covariates obtained from gliders, nets, and continuous underway data logging systems, including the minimum distance from each observation to:
4.1.1 frontal structures (as defined by various sources of hydrographic information)
4.1.2 eddy structures
4.1.3 mean surface and subsurface chlorophyll-a concentration (mgL-1)
4.1.4 mean mesozooplankton and larval fish abundance and diversity, as obtained by continuous vessel or glider-based hydroacoustics surveys, or from net samples.
Due to non-normal density distributions and residuals, a non-parametric approach to investigating physical and biological associations is preferred (e.g., Yen et al. 2006).
Logistic regression is used to test for significant associations between the binomial variable “occurrence” (presence/absence) for each species and habitat characteristics.
Ordered logistic regression is used to investigate the numerical response of seabirds using “presence only” data. The data used in ordered logistic regression approach is re-coded by categorizing “density” into three classes: ‘1’ = low (33 percentile), ‘2’ = intermediate (34-66 percentile), and ‘3’ = high (>66 percentile) density. Since surveys are not conducted at exactly the same time period each year and there is spatial variation in survey effort, to account for both spatial and temporal variability, environmental covariates are assessed after including longitude and latitude (continuous), date (continuous) and year (categorical) in each model.
Transformations of environmental covariates are used to ascertain the functional forms of relationships.
- Hand-held range finder (Heinemann 1981)
- Field computer with DLog data entry program (R.G. Ford Consulting)
- Heinemann, D. 1981. A rangefinder for pelagic bird censusing. Journal of Wildlife Management 45: 489-493.
- Hosmer Jr., D.W. and S. Lemeshow. 2000. Applied Logistic Regression, 2nd Edition. Wiley, New York, NY.
- Yen, P.P.Y., W.J. Sydeman, S.J. Bograd, and K.D. Hyrenbach. 2006. Spring-time distributions of migratory marine birds in the southern California Current: oceanic eddy associations and coastal habitat hotspots over 17 years. Deep-Sea Research Part II 53: 399-418.