The use of Street View maps in urban bird and nest surveys is worth exploring. Using Baidu Street-View (BSV) map, 49338 seamless spherical photos of 2741 sites were collected in Qingdao. Study evaluated the effectiveness and reproducibility of birds and nests using three observation angles and four experimenters. The BSV method and citizen science data were compared to evaluate their spatial distribution consistency. The BSV time machine is used to assess them time dynamics. The environment of the sampling sites with birds or nests was also surveyed information and evaluated the optimal environment. BSV photos can identify bird and nest, and different people had high repeatability on the evaluations. Bird nests selection middle-view BSV images could save 2/3 time used in the checks with >93% precision. Most citizen bird-watching sites overlapped or were close to nest hotspot areas from the BSV method and nests were more prevalent. The BSV time machine is suitable for investigating bird's nests. Nests and birds are more likely to be found in the leafless early spring, on large, traffic-dense coastal streets with complex vertical tree structures, and around tall buildings. For further urban bird research and conservation, this approach provides a pre-experimental and informative supplement.

For making a map of birds and nests distribution in Qingdao city, we used the grid sampling method for collecting BSV images. The grid was generated at a 0.0045° longitude interval and a 0.005° latitude interval on the BSV map of Qingdao, i.e., the grid size was about 500 m * 500 m. There are 2741 suitable sampling sites in total collected in this paper. The collection of BSV images was automatically downloaded by a web crawler developed using Python (version 2.7) to read the latitude and longitude of each site through the BSV image Application Program Interface (API) (Xiao et al., 2021). At each sample point, we collected a total of 18 pictures in 6 horizontal directions (360° rotated by 60° for each viewpoint) and three vertical directions (45° sky-viewing, 0° middle-viewing, and -45° ground-viewing). The nest and bird numbers were counted by the naked eye of the experimenter. The experimenter needs to sift through all BSV images by eye to filter out those with birds or nests for counting. All birds in aerial flight, aerial stay, and ground stay were counted. 

Most of the street maps have been updated several times, and the "time machine" allows us to see two different times of BSV information at the sampling point in 2016 (the first surveying time) and in 2019 (rechecking time for bird and nest observation). In the first survey of 2016, we screened the BSV pictures, including birds and nests. In 2019, we surveyed the BSV sites of 2016. The changes in the number of birds and nests at different times were used as characteristics of their temporal changes by data counting and spatial distribution. Street view information is also obtained here. The land use type in the sampling point was also surveyed in a square of 500m*500m around the sampling point as the center in the top-down Baidu Map. The "cmd program" was used to segment the 500m*500m Baidu map into 100 grids (Fig. A1), facilitating the estimate of the proportion of different land types in each site. We also use Baidu Map's distance tool to measure the distance from each sampling point to the sea and the distance to other water surfaces.

Citizen science data were from the website of the China Birding Records Center (http://www.birdreport.cn/home/search/page.html). Citizens' bird data upload dates were from December 7, 2014, to March 30, 2021, including latitude and longitude coordinates, observation time, bird species, and number. There are many bird enthusiasts uploading bird observation records. We need to filter the data of citizens located in built-up areas.

Python

Arcgis

JMP

Excel

cmd program