See below to download a PDF file

Using Low-level Aerial Surveys to Verify Air Photo Interpretation of Marbled Murrelet Nesting Habitat in Haida Gwaii

Author(s) or contact(s): A. Cober, F.L. Waterhouse, A.E. Burger, A. Donaldson, B. Smart, and P.K. Ott
Source: Coast Forest Region
Subject: Aerial Photography, Birds
Series: Technical Report
Other details:  Published 2012. Hardcopy is available.


Air photo interpretation (API) is widely used in British Columbia for assessing and mapping forest nesting habitat of the Marbled Murrelet (Brachyramphus marmoratus). For strategic land use planning in Haida Gwaii, the entire land base (approximately million ha) has been mapped based on polygons assessed with API. The API method does not, however, detect microsite features in the forest canopy, especially the presence of mossy mats and other potential nest platforms, which are essential for murrelets to nest. This study was therefore undertaken to compare the API habitat quality classifications with those made by low-level helicopter surveys (aerial surveys), which focus on the canopy structure and the presence of potential nest

Both API and aerial surveys use standard protocols, tested in many parts of coastal British Columbia, and use the same six-class habitat-quality system (Nil to Very High). Our comparisons were made from 2004 through 2006 at 191sites clustered in five regions of Haida Gwaii, representative of the common biogeoclimatic zones and subzones. All sites were within forest greater than 140 years old and we did not include areas expected to have Nil habitat quality class. At all sites, assessments were made at two spatial scales: small circular patches (100 m radius; 3.1 ha); and the surrounding polygons of relatively uniform forest (variable in size, but most from 5 to 100 ha). The original API mapping assessed only polygons but for this study we undertook a second, blind post-survey API assessment of both patches and polygons for the study sites. We found no statistically significant difference in the classifications produced by the original API mapping (two observers working independently) and the post-survey API (one observer doing all sites); 87% of the sites were classified the same.

There was no significant difference in the classifications made by the two aerial survey observers for three key habitat features (presence of large trees, potential nest platforms, and moss development). The overall habitat quality classifications made by the two observers were identical at 97.6% of sites for both patches and polygons. Training, experience, and discussion to reach consensus classifications while hovering over the site should provide consistent aerial surveys of murrelet nesting habitat.

The post-survey API showed no significant difference in habitat quality classification for patches and polygons (81% rated identically; n = 190). Similarly, the aerial surveys showed no significant difference in habitat quality between patches and polygons (81% identical; n = 191). Polygons, which can be assessed more rapidly and cheaply than patches by API or aerial surveys, are generally used for large-scale strategic planning and management. Assessment of patches or small forest stands with aerial surveys is useful in finer-scale operational management and selection of critical nesting habitat.

When we pooled data from all five study areas, we found significant predictive relationships between post-survey API and aerial survey classifications at both patch and polygon scales. Where differences existed they were symmetrical (i.e., neither method produced a systematic bias either way). For patches, aerial surveys classified 42% of the sites identically to API, 36% higher, and 22% lower. For polygons, aerial surveys classified 45% of the sites identically, 27% higher, and 28% lower than with API. We developed proportional odds models from the data to predict the probabilities of aerial survey classes from the post-survey and mapping API. The models suggest that the aerial survey classifications would be identical to those from API in 34-60% of sites and within one rank above or below in 82-94% of sites. This proportional odds model can be applied in land use management in Haida Gwaii to show the expected reliability of large-scale forest classifications based on API.

The Strategic Land Use Agreement (SLUA) and Land Use Objectives Order (LUOO) for Haida Gwaii include specific management objectives for the Marbled Murrelet, which are based on retention of areas with the higher-quality nesting habitats (usually simplified to 75% of the combined Class 1 [High] and 2 [Very High] habitats). Our study provides guidance in the reliability and applicability of the API-derived habitat map when setting landscape targets for retention of Marbled Murrelet nesting habitat. Our results are also useful in operational planning to meet those habitat targets when delineating Wildlife Habitat Areas (WHAs) or Forest Reserves. Aerial surveys can confirm habitat suitability and show if adjustments to Forest Reserves are needed to capture suitable nesting habitat; such adjustments are allowed under LUOO. Based on our results, areas classified as Very High should not need field confirmation (e.g., aerial surveys) but areas classified as High or Moderate by API might require such confirmation. Consideration of areas within the Moderate class could, if upgraded by aerial surveys, help achieve the LUOO targets and provide additional flexibility for trade-offs at the operational stage.

Download Technical Report 070 PDF file (574 KB)

To view this document you need the current version of
Adobe Acrobat Reader, available free from the Adobe Web Site.

Updated March 30, 2012