When asked to name the main sources of biodiversity data, researchers might immediately think about large databases such as the Global Biodiversity Information Facility, Ocean Biodiversity Information System or iNaturalist, to name just a few. These databases provide ample information about current distributions of species, but they cannot provide information about species distributions in the longer term, decades before the database records began.

Natural history archives and herbaria are also well recognized as resources for biodiversity information over longer time periods. In this issue, a Review by Daru and Zhigila summarizes how herbarium records can be used to draw inferences about interactions between plants and other organisms (including pollinators, herbivores and pathogens) as well as, importantly, changes in these interactions over time.

But even natural history records and herbaria are somewhat ‘conventional’ data sources used by ecologists and evolutionary biologists. Another potentially valuable approach to understanding biodiversity change over time is to turn to less conventional sources, such as art, books and archaeological records.

A Perspective by Navarro and colleagues in this issue highlights how extracting information from these sources can provide valuable information about biodiversity change over long time periods. One salient example is a case in which oral histories, written records and photographs were integrated to reconstruct 19th and 20th century boundaries between forests and savannahs in southern Guinea — information that contradicted colonial narratives about the timing of deforestation. Another Review in this issue, by Nogué and colleagues, takes a deep dive into the history of human effects on biodiversity in the past; much of the literature they review is drawn from historical sources, including archaeological, palaeontological and sedimentary records.

Navarro et al.’s Perspective outlines seven suggested ways forward to advance this field of research, known as historical ecology. One of their suggestions is to expand and coordinate digitization efforts. Artificial intelligence has a clear part to play in the digitization endeavour. A Perspective in this issue by Silvestro and Pimiento highlights emerging methods for artificial intelligence in palaeontology. Specifically, artificial intelligence has the capacity to accelerate and automate the collection of morphological data from fossil specimens and the literature, as well as empower researchers who are modelling evolutionary processes.

Another of Navarro et al.’s recommendations is to embrace interdisciplinarity and transdisciplinarity; biodiversity scientists should consult experts in history and the social sciences before drawing inferences from art and other historical records. Such bridges — between natural science and social science, and even the arts — are of the utmost importance in an era of biodiversity and conservation science that recognizes the roles of social and political drivers of biodiversity decline.

“Such bridges — between natural science and social science, and even the arts — are of the utmost importance”

Nature Reviews Biodiversity has a scope that is inclusive of these transdisciplinary approaches. For example, a Viewpoint in this issue authored by five political ecologists discusses the importance of sociopolitical worldviews to conservation science.

With biodiversity in decline, studies of contemporary patterns of species distributions and population trends should be complemented by looking back (to a variety of historical sources) and to the side (to other scholarly fields that touch on environmental issues).