Covid-related surge in global wild bird feeding: Implications for biodiversity and human-nature interaction

II. Data a. Google Trends Google is, by far, the most widely used search engine in the world, accounting for 90% of the worldwide search market share over the past ten years [32]. Google Trends data provided by Google supplies an index to show the popularity of terms searched in Google across various regions and languages within a given period. Therefore, Google Trends data can be considered a representative sample of internet search queries globally. Google Trends searches predict behavior change in the stock market [33], tourism [34], suicide [35], and Covid-19 spread [36]. For bird feeding specifically, Google searches for bird feeding correlate with bird feeding behavior: Brock et al. [28] use data from Cornell’s Project FeederWatch and find an increase in bird feeding after Covid-19 lockdowns in the U.S. that mirrors a similar increase in Google searches for bird feeding terms. We use Google search data to assess whether residents are interested in bird feeding, as opposed to measuring the intensity of feeding. We used two types of data provided by Google Trends to explore global bird engagement. First, we used a panel (01/01/2019 to 05/31/2020) of weekly indexed search frequency data to examine the change in search interests over time because of the Covid-19 lockdowns. We observed the data in a nation for each week (e.g., the search frequency of a search term in the U.K. in week 30 of the year 2019). Our study period follows Rousseau and Deschacht [37], a study that explores public awareness of nature and the environment during the COVID-19 cri-

sis using Google Trends data. The relative search intensity was calculated as the number of daily searches for the search topic relative to all other search queries in the given time and region. The Google Trends index represents the relative search intensity of a search topic over the requested time period in a geographical area relative to the highest point for the given region and time. It is scaled from 0 to 100, where 100 is the week with the most searches for that topic. If the search volume is very low throughout the period, Google Trends does not report data for that geographical area for the whole period. Second, we used cross-sectional data to explore the search interest of a search topic by nation during the same time period (01/01/2019 to 05/31/2020). The data include countries with lower search volume, which is recommended by Google as a method to include smaller countries. This index was also calculated from 0 to 100, where 100 represents the location where the search topic was most popular during the selected time frame. A high index means a higher relative search intensity (proportion of all queries) but not a higher absolute number of queries in a given region. Countries with very low search intensity do not receive a Google Trends index. Google does not report the threshold proportion of all queries at which they stop reporting an index. For the cross-sectional data, we treat the actual search intensity as an unobserved, latent variable that is only observed when it crosses this threshold. We then re-

code the Google Trends index to reflect the existence of search interest, with values that have a reported Google Trends index receiving a 1 and values without an index receiving a 0. Topics on Google Trends were a group of terms that shared the same concept in any lan- guage, which is useful for accommodating different languages used worldwide. To create our sample for the global analysis, we focused on three bird engagement related topics: "bird feeder", "bird food", and "bird bath". Bird feeders, bird food, and bird baths are popular bird- watching equipment [38]. Households use such tools to attract more bird species to visit their home and gardens [39]. The set of countries and territories with sufficient search volume var- ied across topics. During the study period, 115 countries (60% of all countries) and territories had sufficient volume for the "bird feeder" topic, 79 for the "bird food" topic, and 56 for the "bird bath" topic. The counts are different for the panel analysis. The exact number of coun- tries included in each analysis are reported in Table A3 in S1 File. PLOS ONE Covid-related surge in global wild bird feeding PLOS ONE | https://doi.org/10.1371/journal.pone.0287116 August 2, 2023 3 / 15