Just Google it: assessing the use of Google Images to describe geographical variation in visible traits of organisms

Abstract

Summary Describing spatial patterns of phenotypic traits can be important for evolutionary and ecological studies. However, traditional approaches, such as fieldwork, can be time‐consuming and expensive. Information technologies, such as Internet search engines, could facilitate the collection of these data. Google Images is one such technology that might offer an opportunity to rapidly collect information on spatial patterns of phenotypic traits. We investigated the use of Google Images in extracting data on geographical variation in phenotypic traits visible from photographs. We compared the distribution of visual traits obtained from Google Images with four previous studies: colour morphs of black bear (Ursus americanus); colouration and spottiness in barn owl (Tyto alba); colour morphs of black sparrowhawk (Accipiter melanoleucus) and the distribution of hooded (Corvus corone) and carrion crows (Corvus cornix) across their European hybrid zone. Additionally, we develop and present a web application (Morphic), which facilitates the human data capture process of this method. We found good agreement between fieldwork data and Google Images data across all studies. Indeed, there was strong agreement between the data obtained from the original study and from the Google Images method for the colour morphs of black bear (R2 = 80%) and for two barn owl plumage traits (R2 = 64% and 53%). Our approach also successfully matched the clinal variation of black sparrowhawks morphs across South Africa. Our method also gave a good agreement between the distribution of hooded and carrion crows (with 86% placed on the correct side of the hybrid zone line). Our results suggest that this method can work well for visible traits of common and widespread species that are objective, binary, and easy to see irrespective of angle. The Google Images method is cost‐effective and rapid and can be used with some confidence when investigating patterns of geographical variation, as well as a range of other applications. In many cases, it could therefore supplement or replace fieldwork.