75 years of dryland science: Trends and gaps in arid ecology literature

Abstract

Growth in the publication of scientific articles is occurring at an exponential rate, prompting a growing need to synthesise information in a timely manner to combat urgent environmental problems and guide future research. Here, we undertake a topic analysis of dryland literature over the last 75 years (8218 articles) to identify areas in arid ecology that are well studied and topics that are emerging. Four topics-wetlands, mammal ecology, litter decomposition and spatial modelling, were identified as 'hot topics' that showed higher than average growth in publications from 1940 to 2015. Five topics-remote sensing, climate, habitat and spatial, agriculture and soils-microbes, were identified as 'cold topics', with lower than average growth over the survey period, but higher than average numbers of publications. Topics in arid ecology clustered into seven broad groups on word-based similarity. These groups ranged from mammal ecology and population genetics, broad-scale management and ecosystem modelling, plant ecology, agriculture and ecophysiology, to populations and paleoclimate. These patterns may reflect trends in the field of ecology more broadly. We also identified two broad research gaps in arid ecology: population genetics, and habitat and spatial research. Collaborations between population genetics and ecologists and investigations of ecological processes across spatial scales would contribute profitably to the advancement of arid ecology and to ecology more broadly.

Fig 1. Growth in research articles per topic.

The frequency per year of research articles published for each topic area in the global dryland literature from 1940 to 2015. Topics were identified by Latent Dirichlet Allocation using abstracts from 8218 articles.

Fig 2. Hot topics in global dryland research.

Identified hot (red circles) and cold (blue circles) topics in global dryland literature from Poisson Generalised Linear Mixed Model slopes and intercepts. Topics with positive random intercepts can be interpreted as having higher than average numbers of articles written about them, and topics with random slopes greater than the fixed-effect mean (± 95% confidence interval) have higher than average growth in publications in the time period analysed (1940-2015). Topics with average growth over the study period are shown with black circles. See Fig 3 and S1 Table for identification of topic numbers.

Fig 3. Topic similarity in global dryland research.

Topic similarity in global dryland literature calculated using the Euclidean distance between each pair of topics using a Latent Dirichlet Allocation matrix of the weight of each word within each topic.

Fig 4. Research gaps in global dryland literature.

Research gap distance matrix heat-map (red = high, clear = low) on global dryland literature. The greater the metric the higher the dissimilarity between topics (i.e. topics that both contain different sets of words and topics that rarely co-occur in the same article).