Trend and Annual Index Information

Population trend estimates are the simplest measure of population change over a time period. They are expressed as a percent change per year, and in these Tables they are presented for many regions (or species) and two time periods.

Here is the header and one line of data for a "typical" trend output:

1. Regions (or Species)

This column defines the group for which the trend is defined. Regions presented in this analysis are states, Provinces, Bird Conservation Regions (BCR), BBS Regions, Canada, US (lower 48 states), and Survey-wide (excluding data from Alaska and NF).

Bird Conservation Regions are used as the primary physiographic strata for the analysis. BCRs were developed to provide a coherent spatial framework for conservation planning efforts under the North American Bird Conservation Initiative (NABCI) . These regions were derived from existing physiographic strata (see map and explanation at the Bird Studies Canada BCR site. BBS regions are groups of physiographic strata. The Eastern Region (EA) is composed of strata east of the Mississippi River in the lower 48 US, but also includes forested strata in central Canada. The Central Region (CE) extends from the Mississippi River to the edge of the Rocky Mountains. The Western Region includes all strata to the west of the Central and Eastern Regions. Note that Alaska, Newfoundland, Yukon, and areas clearly to the north of the survey area are not included in any regional summary. The BBS regions were first defined in a paper by Danny Bystrak (Bystrak, D. 1981. The North American Breeding Bird Survey. Studies in Avian Biology 6:34-41, Figure 1). The present regions cover similar areas, but exact boundaries are based on Bird Conservation Regions rather than the BBS physiographic regions used in earlier analyses. This map shows the regions as groups of BCRs (Green=Eastern, Yellow=Central, Red=Western).

2. Sample Size

This number represents the number of survey routes on which the species was encountered during the long-term (1966-2009) interval. We note that earlier summaries of BBS data defined sample size slightly differently, as the number of routes for which trends could be calculated.

3. 1966-2009 Trend Estimates

These are trend estimates calculated as the ratio of endpoints from the annual indices produced in the hierarchical model analysis (See 8, below). They are presented as a yearly percentage change.

4. 1966-2009 Credible Interval for Trend Estimate

These values represent the 2.5% and 97.5% percentiles of the posterior distribution of trend estimates. These values replace the precision estimates and p values presented in earlier analyses. If the credible interval does not contain 0, the result could be judged significant, and in these tables significant results are indicated by red (for declining populations) or blue (for increasing populations) print.

5. 1999-2009 Trend Estimates

These are trend estimates calculated as the ratio of annual indexes from 2009 and 1999, presented as a % change/year. This 10 year interval describes recent patterns of population change.

6. 1999-2009 Credible Interval for Trend Estimate

These values represent the 2.5% and 97.5% percentiles of the posterior distribution of trend estimates. If the credible interval does not contain 0, the result could be judged significant.

7. Relative Abundance

The value is the annual index for the region from year 22. The quantity differs from the relative abundance measure provided in earlier analyses as it is model-based, produced as part of the hierarchical model analysis. As such, it is a predicted index for year 22, that is adjusted for observer and other effects. Earlier abundance measures were simple route averages within state-strata areas, area-weighted to get regional means.

8. Annual Indexes

Annual indexes are linked to the region (or species) names. Clicking on the name will provide a graphic of the annual indices and their credible intervals. Note that the graphics were produced to ensure that the annual indexes were visible with axes that permit observation of pattern in the indices. This scaling occasionally cuts off extreme values in the upper credible intervals.