https://doi.org/10.5061/dryad.83bk3jb2r

Description of the data and file structure

Our narrative synthesis was supplemented with references and guided by literature drawn from text-mining and topic modelling of 9,856 peer-reviewed papers. We searched the Clarivate Analytics Web of Science titles containing the following word combination:

(“older” OR “aging” OR “ageing” OR “growing old” OR “longevity” OR “lifespan” OR “old-growth”) AND ("vertebrates" OR “animals” OR “fish*” OR “mammals” OR “reptiles” OR “birds” OR “frogs” OR “crustaceans” OR “mollusks” OR “sponges” or “coral” or “invertebrates” OR “insects”) AND (“ecology” OR “life-histor*” OR “pace-of-life”  OR “behav*” OR “reproduct*” OR “knowledge” OR “cultural transmission” OR “food web” OR “diet” OR “climate change” OR “global change” OR “resilience” OR “physiology” OR “conservation” OR “fisheries” OR “fishing” OR “wildlife” OR “hunting” OR “harvest”)

NOT dog OR cat OR mouse OR mice OR drosophila OR "fruit fly" OR cattle OR cow OR chicken OR rat*

We collected all research papers (hereafter “articles”) from 1990 up to September 28 2023, resulting in a return of 9,856 articles. We used text from title, keywords and abstract to form the article content, and every article was tokenized (i.e., the process of obtaining individual words—also known as unigrams—from sentences). We removed meaningless terms including stop words (for example: the, or, and, which), numbers and punctuation. Minimum word length was set to three characters. The remaining words were stemmed (reduced to their base or root form, for example diverging and divergent become diverg). We then removed very rare (n < 20) and very common terms (n > 6,000) that provide little information content, resulting in 741,328 entries of 6,386 unique terms. The text-mining and topic model originally identified 15 emergent topics which are listed in the dataset. However, nine of the emergent topics such as those related to human health, animal meat production and methods of age determination were not directly related to the scope of this review were not reviewed.

Latent Dirichlet Allocation (LDA) modelling was used to identify common topics reported in our dataset. LDA identifies sets of co-occurring words that are more frequently presented within the same linguistic context than expected by chance alone. Outcomes from the LDA include a list of the most common words and their topic probabilities for each article which are the numerical values provided in the dataset

Files and variables

File: topic_model_results.csv

Description: 

Variables

• Paper.ID: Author, year, journal
• Article.Title:
• 1: topic
• 2: topic
• 3: topic
• 4: topic
• 5: topic
• 6:topic
• 7:topic
• 8:topic
• 9:topic
• 10:topic
• 11:topic
• 12:topic
• 13:topic
• 14:topic
• 15:topic

Access information

Data was derived from the following sources:

• Clarivate Analytics Web of Science