Evolving trends in vertebrate palaeontology (2013–2022): a bibliometric analysis using DeepBone and Web of Science databases

ABSTRACT

Vertebrate palaeontology, essential for understanding the evolutionary processes of backboned animals, has witnessed substantial increase in research activity and publications over the past decade. This bibliometric analysis utilising the data from the DeepBone database, which includes 9,255 literature entries, with 8,259 overlapping entries from the Web of Science, to capture these changes. Our analysis reveals a continuous growth in publication volumes over the decade, alongside a marked expansion in interdisciplinary collaborations and the integration of innovative technological methodologies. The contribution of key regions such as the U.S.A., UK, Argentina, and China underscores the global collaboration in vertebrate palaeontology. The EU, despite its foundational economic and political union focus, stands out as a key contributor in vertebrate palaeontology, showing its extensive influence on scientific collaboration and research funding. A thematic exploration using keyword analysis and Latent Dirichlet Allocation (LDA) analysis identifies prominent research themes, highlighting the field’s towards integrated research approaches. This study offers a quantitative overview of the recent decade in vertebrate palaeontology, providing insights into dynamic trends, major contributions, and future research directions in the field.

KEYWORDS:

• Scientific trends
• vertebrate palaeontology
• bibliometric analysis
• DeepBone database
• Web of Science

Introduction

Vertebrate palaeontology, a branch of palaeontology, is pivotal in understanding biological evolution and palaeoecology of backboned animals to which humankind belongs. As the volume of publications in this field increases, it becomes essential to analyse these works not only to gain historical perspective but also to highlights potential future directions. This paper employs a bibliometric analysis to provide a quantitative evaluation of current trends in vertebrate palaeontology, offering a structured roadmap for future investigations.

French scientist George Cuvier initiated the study of vertebrate palaeontology during the period of 1796–1812, when he used comparative anatomy to demonstrate that the Mammuthus (mammoth) and Mammut (mastodon) bones found near Paris belonged to extinct species (Cuvier 1812; Gregory 1979). The field gained momentum with Swiss naturalist Louis Agassiz’s (1807–1873) Recherches sur les poissons fossiles in 1833–1843 (Agassiz 1833–1843) and further expanded with Charles Darwin’s (1809–1882) On the Origin of Species in 1859, which provided a theoretical framework focusing on the evolutionary relationship and process between fossils and extant organisms (Darwin 1859). The establishment of The Society of Vertebrate Palaeontology in 1940 and the Journal of Vertebrate Palaeontology in 1980 marked rapid development in this field (Cain 1990). With the advent of molecular biology and other new technologies in the 1970s, vertebrate palaeontology has flourished. Although there have been studies reviewing different groups, regions, and techniques of vertebrate palaeontology (Lister and Sher 2001; Lucas 2014; Otero et al. 2020), our work, distinct from traditional reviews, which can be somewhat subjective (Guan et al. 2022), utilises advanced bibliometric tools and databases to offer a data-driven, multi-analysis of the field’s progression over the past decade.

Bibliometrics, as a quantitative analysis of references, has growth extensively in modern biological sciences, becoming a recognised tool in research analytics (Mishra et al. 2021; He et al. 2022), resulting in software like VOSviewer, CiteSpace, and R (Bibliometrix package). Applying bibliometrics to the analysis of vertebrate palaeontology research can help researchers accurately grasp the current status and trends of vertebrate palaeontology, providing a basis for understanding what underlies its development, and offering valuable insights such as identifying potential research gaps, forecasting emerging trends, and understanding regional contributions to the field. Our study leverages the DeepBone database to present one of the most exhaustive bibliometric evaluations of vertebrate palaeontology, mapping out pivotal themes, key institutions, and geographical regions of significant academic impact, thereby elucidating the current landscape and future potentialities of vertebrate palaeontological research.

The completeness and accuracy of the references are vital for ensuring that bibliometrics research objectively reflects the development characteristics of a discipline (Hayashi and Mitchell 2019). Most bibliometric research mainly uses keyword searches in various databases to gather references data. Therefore, the rationality of keyword selection is key to the accuracy of bibliometric analysis results. However, keyword searches in databases often face limitations, potentially resulting in overly broad or narrow scopes. This can lead to either the inclusion of irrelevant references or the omission of pertinent ones, impacting the accuracy and comprehensiveness of the bibliometric analysis. In 2019, the development of the DeepBone Database was initiated by experts in vertebrate palaeontology, aiming to create the most comprehensive and high-quality database of vertebrate palaeontology on the internet. A dedicated team was formed for data entry, and over the years, DeepBone database (www.DeepBone.org) has become the most exhaustive database on vertebrate paleontology on the internet. Now, the DeepBone database provides a comprehensive references source and provides a solid data foundation for analysing the progress of vertebrate palaeontology research based on bibliometrics (Pan and Zhu 2019, 2020).

To the best of our knowledge, the present study represents the first comprehensive bibliometric analysis of vertebrate palaeontology. By leveraging this specialised and exhaustive data source, our work provides uniquely detailed insights into the research trends and future directions of the field during the last ten years.

Materials and methods

Data source

This study is based on the DeepBone database, comprising a total of 9,255 references as of 27 April 2023, which includes limited numbers of conference abstracts as catalogued in the Web of Science (WOS). DeepBone was chosen for its comprehensive coverage and accuracy in vertebrate palaeontology. Instead of relying on direct keyword-based searches in WOS, we utilised the DeepBone database as the primary reference to identify and retrieve relevant articles from WOS. This approach ensures specific alignment with vertebrate palaeontology and overcoming the limitations of broader keyword-based searches. Out of the references from DeepBone, 8,259 articles (89.2%) were matched and imported from WOS (Figure 1). WOS was selected due to its status as a reputable scientific references database, covering a broad spectrum of scientific disciplines, WOS offers the best compatibility as a data source and is frequently used in bibliometric studies (Niu et al. 2021). The remaining 964 articles (10.2%) were not found in the WOS, but relevant information for 181 of these articles (1.96%) was located in the China National Knowledge Infrastructure database (CNKI, www.cnki.net/). The remaining 783 articles (8.24%) were either obtained through manual searches of specialised databases or directly from the publication sources.

Figure 1. Flow diagram showing document selection and data analysis process. This figure illustrates the process of references selection from the DeepBone database, and the workflow of data analysis using different softwares.

For visualisation analyses, we used the WOS data, as the analytical software employed is optimised for WOS data structures and export formats, ensuring the integrity and compatibility necessary for our visualisation tools, which rely on WOS’s consistent and standardised metadata.

Data analysis method

Our analysis encompasses the quantitative trajectory of publication volume over time, the geographical distribution of research contributions, the academic and research entities involved, preferred dissemination channels (journals), and sources receiving significant scholarly attention (highly cited journals). Additionally, we explored keywords to reveal development trends in vertebrate palaeontology, relationships between countries and institutions, publication status of journals, research hotspots, and systematically described the current status and progress of vertebrate palaeontology research in the last decade. This approach allowed us to offer a systematic, quantitative analysis of the evolution of vertebrate palaeontology literature from 2013 to 2022, a period capturing a decade of profound changes in the realm of scientific research, driven by technological advancements and global events.

VOSviewer (version 1.6.18) was utilised to construct collaboration networks due to its advanced algorithm for spatial visualisation (Van and Waltman 2010). CiteSpace was selected for its keyword analysis functionality optimised for references review (Chen 2016). Bibliometrix provided publication statistics aligned with our research questions (Aria and Cuccurullo 2017). The complementary capabilities of these tools enabled a multi-faceted bibliometric investigation.

The bibliometric analysis was conducted using a four-step process, each step employing a specific software tool for different aspects of the analysis (Figure 1).

In the first step, OriginPro 2021 (64-bit, version 9.8.0.200) was employed to illustrate several distinct trends within vertebrate palaeontology (Chemical Engineering group 2021): 1) the annual number of publications, 2) the shifts in the percentage of articles published in the prominent CNS journals (Cell, Nature, Science), and 3) the patterns in the frequency and characteristics of newly described taxa over time.

The second stage of investigation focused on pinpointing the countries and institutions that were most prolific in their contributions. Here, VOSviewer software was utilised for creating visual mappings of the top 20 contributing countries and the online analysis Charticulator (https://charticulator.com/) was employed to elucidate the network of international collaborations (Arruda et al. 2022).

In the third step of our analysis to further dissect the collaborative framework we imported the net, clu, and vec files generated by VOSviewer into Pajek software (version 5.16), which facilitated the construction of a network map showcasing the interconnections between the top 30 contributing institutions, and keywords cluster to showcase the hotspots of vertebrate palaeontology studies over the past decade and their interrelations (Mrvar and Batagelj 2016). Furthermore, we utilised CiteSpace software to analyse the top ten journals in terms of publication volume and citation count (Chen 2016), we also searched the WOS for the 2022 impact factor of these journals, and the Cited Half-life (years) for the top ten journals in terms of publication volume as well as to explore their influence and centrality in the field. Using CiteSpace, we exported the top ten most cited vertebrate palaeontology articles of the last decade to analyse the research focus of these highly cited papers. Despite CiteSpace (version 6.2.2) being particularly noted for its keyword clustering and trend analysis (Chen 2016), it also offers sophisticated tools for journal metric evaluations, making it an appropriate choice for this dimension of our study.

The fourth step utilised the Bibliometrix and ggplot2 packages (version 3.4.3) in R 4.3.1 (R Core Team 2022) for a fundamental bibliometric analysis and visualisation (Aria and Cuccurullo 2017). The Bibliometrix package provided an in-depth analysis of the primary research themes in vertebrate palaeontology, while ggplot2 was employed to craft insightful keyword distribution maps and a cumulative keyword trend visualisation, thus offering a macroscopic perspective on the evolving discourse within the field (Aria and Cuccurullo 2017). Subsequently, Latent Dirichlet Allocation (LDA) was employed to uncover the underlying thematic structure within the corpus of vertebrate palaeontology abstracts sourced from the Web of Science database. Data for LDA was derived from the previously mentioned WoS-exported Excel file (Blei et al. 2003; Jelodar et al. 2019), ensuring consistency with prior analyses. Abstracts missing from the Excel dataset were excluded from the analysis. To maintain processing consistency, each abstract was normalised to lowercase text. A term frequency matrix was constructed using the 5000 most frequently used terms, excluding common English stop words. An LDA model with five components was then trained on the term frequency matrix. This training allowed the identification of latent topics by associating terms that frequently co-occur across the dataset. A visualisation of the topic modelling was prepared using the pyLDAvis library (version 3.4.1, https://pypi.org/project/pyLDAvis/), providing an interactive view of the term distributions across the identified topics and the distribution of topics across the documents (Sievert and Shirley 2014). The visualisation was saved as an HTML file, allowing for in-depth exploration and interpretation of the thematic structure. This LDA-based analysis facilitated a nuanced understanding of the prevailing trends within the field, revealing valuable insights into the focal points of vertebrate palaeontology research. The complete code used for the LDA analysis can be found in the supplemental materials section, providing a comprehensive overview of the method and results.

For the convenience of this study, we have categorised vertebrate palaeontology into five groups: fishes, amphibians, reptiles, birds, and mammals. However, we acknowledge that defining these categories is a complex task due to evolutionary continuities and the existence of species that challenge traditional clade definitions (Benton 2014). Our categorisations largely rely on widely accepted classifications though we provide specific rationales for species that are typically at the centre of taxonomic debate. For instance, we classify Tiktaalik as a fish and feathered dinosaurs as reptiles. It is important to clarify that our categorisations are not definitive and are used here for the functional convenience of this study.

Results

Annual distribution of publications

Between 2013 and 2022, a total of 9,255 papers were published in the field of vertebrate palaeontology. The overall trend shows a yearly increase, which can be divided into three stages: a steady growth phase (2013–2015), during which the annual number of publications was relatively low but showed a slow increase; a rapid growth phase (2015–2019), during which the annual number of publications increased rapidly, reaching a peak in 2019 with 1,444 papers; and a fluctuating decline phase (2019–2022), during which the annual number of publications overall decreased, with minor fluctuations in between (Figure 2a).

Figure 2. Overview of annual publications in vertebrate palaeontology during 2013–2022. a, Number of annual publications; b, Proportion of CNS (Cell, Nature, Science) papers; c, Proportion of new taxon papers; d, Average number of authors per paper; e, Average number of nations per paper; f, Proportion of fishes papers; g, Proportion of amphibians papers; h, Proportion of reptiles papers; i, Proportion of birds papers; j, Proportion of mammals papers. Detailed table of proportions of studies on different vertebrate fossil taxa over the past decade see supplement.

The average annual proportion of CNS (Cell, Nature, Science) papers has shown a low increase over ten years, with a substantial growth in the last year, rather than stabilising, with variations indicating an inconsistent annual growth rate, as depicted in Figure 2(b,c). Average number of authors per paper and the average number of nations per paper have both shown a significant increase over ten years (Figure 2d,e). Over the last decade, reptilian studies have consistently attracted the most attention within vertebrate palaeontology, averaging 38.38% of the annual publication share despite a modest upward trajectory. Mammalian research, holding the second position, constitutes an average annual proportion of 30.00%, also with a gentle upward trajectory. Ichthyological studies are the third most frequent, with 15.78% on average, and are depicted with a gradual downward pattern. Avian and amphibian research, although less prevalent with average annual proportions of 6.26% and 3.58%, respectively, show gentle negative trends as illustrated in Figure 2(f,j), which may be related to factors such as the lower likelihood of fossilisation due to bone fragility, as discussed later.

The analysis identified the main research directions in vertebrate palaeontology by examining the content of published papers, regardless of the journals’ classifications. It revealed that the prominent areas of focus over the past decade were Palaeontology (3649, 44.18%), Multidisciplinary Sciences (2084, 25.23%), and Evolutionary Biology (1083, 13.11%). This analysis focused on the actual content of the research rather than the subject categories assigned by the journals, acknowledging that categorisation in journals may not always align with the actual research themes. For instance, papers centrally dealing with palaeontological topics may be catalogued under ‘Multidisciplinary Sciences’ by a journal, reflecting its broader scope but not necessarily the specific content of the paper. This assessment highlights a potential discrepancy between journal categories and research content, underscoring the importance of analysing the papers’ actual subjects to understand the true trends in vertebrate palaeontology (Table 1).

Table 1. Primary research fields in vertebrate palaeontology from 2013 to 2022. This table summarises the primary research fields related to vertebrate palaeontology between 2013–2022 based on the subject categories of the published papers, with the frequency and proportion of papers in each field.

Study field	Frequency	Proportion
Paleontology	3649	44.18%
Multidisciplinary Sciences	2084	25.23%
Evolutionary Biology	1083	13.11%
Biology	850	10.29%
Geosciences, Multidisciplinary	756	9.15%
Zoology	582	7.05%
Geology	534	6.47%
Ecology	510	6.18%
Geography, Physical	253	3.06%
Anthropology	238	2.88%
Anatomy & Morphology	234	2.83%
Biodiversity Conservation	220	2.66%
Biochemistry & Molecular Biology	134	1.62%
Cell Biology	114	1.38%
Genetics & Heredity	73	0.88%

Note: A references may belong to multiple disciplines according to the WOS discipline classification, so the total percentage of disciplinary directions is greater than 100%.

Contributions of countries and institutions

In the decade from 2013 to 2022, vertebrate palaeontology research involved 126 countries, based on the affiliation of the authors. Alongside individual European countries like the United Kingdom (1492, 18.07%), and Germany (1239, 15.00%), the European Union (EU, 2441, 51.64%, from 2014 to 2019), as a significant economic and political entity, is also compared to research giants such as the U.S.A. (3061, 37.06%) in terms of research co-occurrence frequency. China (989, 11.97%), Argentina (958, 11.60%), France (932, 11.28%), Spain (779, 9.43%), Canada (728, 8.81%), Australia (634, 7.68%), and Brazil (556, 6.73%) also had relatively high co-occurrence frequencies. Collaborations between the United States and the United Kingdom, Germany, and China were particularly strong, as was the collaborative relationship between the United Kingdom and Germany (Figure 3a,b).

Figure 3. Collaboration networks among the top 20 most contributive entities in vertebrate palaeontology research over the last decade. (a) The chord diagram illustrates the partnership dynamics between these countries. The size of each segment on the outer circle represents the total number of publications a country has contributed, with colour intensity indicating the quantity, ranging from blue (fewest) to red (most). The interconnecting ribbons illustrate the collaboration strength, with ribbon thickness corresponding to the number of co-authored papers. (b) This diagram provides an overview of collaborative interactions within the European region, including both EU member states and non-EU contributors, during 2014–2019. This period encompasses significant events such as Croatia’s accession to the EU and precedes the UK’s Brexit transition, highlighting the evolving nature of research collaborations in Europe.

In the past ten years, 545 institutions have participated in vertebrate palaeontology research. In cases of authors affiliated with multiple institutions, each was counted separately in our analysis. The Chinese Academy of Sciences (CAS) ranked first, with a co-occurrence frequency of 673, accounting for 8.15% of all articles. The National Scientific and Technical Research Council of Argentina (Consejo Nacional de Investigaciones Científicas y Técnicas) and Russian Academy of Sciences had co-occurrence frequencies of 433 (5.24%) and 377 (4.56%), ranking second and third, respectively. This was followed by the Natural History Museum, London (376, 4.55%), American Museum of Natural History (294, 3.56%), University of the Chinese Academy of Sciences (256, 3.10%), University of the Witwatersrand, South Africa (256, 3.10%), University of Bristol (237, 2.87%), University of Alberta (228, 2.76%), and Smithsonian Institution (216, 2.62%). Overall, the top 30 institutions in terms of co-occurrence frequencies had a high level of collaboration, with CAS and the University of the Chinese Academy of Sciences and the American Museum of Natural History having the closest ties (Figure 4).

Figure 4. Partnerships of the top 30 institutions with the highest contribution to vertebrate palaeontology study from 2013 to 2022. This network graph shows the collaboration between the top 30 institutions contributing to vertebrate palaeontology research from 2013–2022. Node size indicates number of publications, line thickness indicates collaboration frequency.

Key publications and highly cited journals

Between 2013 and 2022, research in vertebrate palaeontology was primarily published in 302 journals or monographs. The top ten journals by volume are listed in Table 2. Journal of Vertebrate Palaeontology is ranked first, accounting for 8.65% of all articles with 714 publications. PeerJ (412, 4.99%) and Historical Biology (389, 4.71%) are ranked second and third, respectively. Among the top ten journals by volume, there were six specialised and four comprehensive journals, with an average 2022 impact factor of 2.66, a maximum impact factor of 4.60 and a minimum impact factor of 1.40, average 2022 Cited Half-life of 7.43, a maximum Cited Half-life of 13.90 and a minimum Cited Half-life of 4.00 (Table 2).

Table 2. Top 10 journals in publication and Co-citation for vertebrate palaeontology studies from 2013 to 2022.

Rank	Journals	Frequency	IF(2022)	Cited Half-life (years)	Co-cited Journal	Co-citations	IF(2022)
1	Journal of Vertebrate Paleontology	714	1.4	13.9	Journal of Vertebrate Paleontology	5936	1.4
2	PeerJ	412	2.7	4.1	PLoS One	5066	3.7
3	Historical Biology	389	1.4	4	Nature	4762	64.8
4	Scientific Reports	375	4.6	4.5	Science	4237	56.8
5	Cretaceous Research	372	2.1	6.9	Zoological Journal of the Linnean Society	4028	2.8
6	PLoS One	367	3.7	7.3	Palaeontology	3620	2.6
7	Journal of Systematic Palaeontology	235	2.6	6.8	Proceedings of the National Academy of Sciences of the United States of America	3478	11.1
8	Acta Palaeontologica Polonica	208	1.8	12.3	Proceedings of The Royal Society B-Biological Sciences	3377	4.7
9	Zoological Journal of the Linnean Society	203	2.8	10.3	Palaeogeography Palaeoclimatology Palaeoecology	3364	3.0
10	Royal Society Open Science	156	3.5	4.2	Acta Palaeontologica Polonica	3090	1.8

In the past decade, Journal of Vertebrate Palaeontology have been the most cited journal in vertebrate palaeontology research, with 5936 citations, followed by PLoS One (5066) and Nature (4762). The top ten most cited journals included six specialised and four comprehensive journals, with an average 2022 impact factor of 15.27, a maximum impact factor of 64.8 and a minimum impact factor of 1.4 (Table 2).

The citation count within this study’s corpus indicates the article’s relevance to the field, revealing the frequency with which the article is referenced among the collected body of work. Among the top ten cited articles, five targeted dinosaurs, three focused on the construction of phylogenetic trees, one targeted mammals, and one was a stratigraphic study. Nature published three of these highly influential papers. The journals where these seminal articles appeared boasted an average impact factor of 27.99 in 2022, underscoring their prominence in the scientific community (Table 3).

Table 3. Top 10 Co-cited articles in vertebrate palaeontology from 2013 to 2022. This table lists the top 10 journals ranked by total publication number and co-citation number related to vertebrate palaeontology from 2013–2022, along with the 2022 journal impact factor (IF).

Rank	Co-cited references	Journal	Year	Co-citation	IF(2022)
1	TNT version 1.5, including a full implementation of phylogenetic morphometrics	Cladistics	2016	459	3.6
2	The phylogenetic relationships of basal archosauromorphs, with an emphasis on the systematics of proterosuchian archosauriforms	PeerJ	2016	130	2.7
3	A comprehensive phylogeny of birds (Aves) using targeted next-generation DNA sequencing	Nature	2015	98	64.8
4	A new hypothesis of dinosaur relationships and early dinosaur evolution	Nature	2017	90	64.8
5	Weighted parsimony outperforms other methods of phylogenetic inference under models appropriate for morphology	Cladistics	2018	87	3.6
6	Posterior summarisation in Bayesian phylogenetics using Tracer 1.7	Systematic biology	2018	79	6.5
7	The earliest bird-line archosaurs and the assembly of the dinosaur body plan	Nature	2017	78	64.8
8	The placental mammal ancestor and the post – K-Pg radiation of placentals	Science	2013	71	56.8
9	A unique Late Triassic dinosauromorph assemblage reveals dinosaur ancestral anatomy and diet	Current Biology	2016	70	9.2
10	The ICS international chronostratigraphic chart	Episodes	2013	69	3.1

Frequency and emergence of keywords

In this bibliometric analysis, the keyword analysis is based on the keywords provided by authors in their articles. This approach ensures a focused examination of the topics deemed most relevant by researchers in the field. Figure 5 demonstrates the keyword distribution (frequency of keywords) and overlapped keyword distribution (emergence of keywords). The distribution of keywords reflects the research hotspots in vertebrate palaeontology over the past decade, and it also indicates the frequency of a specific keyword in a given year. The overlapped keyword distribution allows for the observation of early-established keywords through earlier colour changes, and rapidly developing keywords through quicker colour changes. This can highlight keywords that may not have a high overall frequency but appear concentrated in certain years, thereby providing a more comprehensive and accurate portrayal of the hotspots and frontiers in vertebrate palaeontology research at different periods (Song et al. 2022). Between 2013 and 2022, the most frequently occurring keywords in vertebrate palaeontology research were ‘evolution’ (2147), ‘phylogeny’ (816), and ‘origin’ (729). In addition, ‘morphology’ (680), ‘anatomy’ (557), ‘mammals’ (555), ‘phylogenetic tree’ (395), ‘diversity’ (384), ‘basin’ (372), and ‘skull’ (366) were also high-frequency keywords. Moreover, keywords such as ‘China’, ‘birds’, and ‘early evolution’ have consistently appeared at high frequencies since 2014 (Figure 5). The prevalence of keywords such as ‘evolution’ and ‘phylogeny’ underscores the field’s continued focus on elucidating evolutionary relationships. Meanwhile, frequent terminology like ‘morphology’ and ‘anatomy’ reveals an ongoing focus on anatomical and morphological analyses as a cornerstone of vertebrate palaeontology. Furthermore, regular mentions of ‘mammals’ and ‘birds’ underscore the research significance of these taxa.

Figure 5. The temporal distribution and prominence of keywords in vertebrate palaeontology from 2013 to 2022. The left heatmap displays the annual frequency of keywords, with years on the horizontal axis and keywords on the vertical axis. The intensity of colour indicates keyword frequency, with warmer colours representing higher occurrences. The right heatmap shows the cumulative distribution of these keywords over the decade. In this visualisation, the intensity of warmer colours corresponds to the cumulative prominence of keywords. Together, these heatmaps elucidate the individual and overall significance of topics over time within the field.

The keywords cluster network reveals 13 main clusters, highlighting core research themes in paleovertebrate studies such as ‘phylogeny’, ‘taxonomy’, and ‘evolution’, alongside prominent geologic periods like ‘Cretaceous’ and ‘Miocene’. The map also features significant associations between ‘evolution’ and key concepts like ‘speciation’ and ‘extinction’, indicating the central role of evolutionary studies. Additionally, ‘phylogeny’ frequently co-occurs with ‘taxonomy’, reflecting their interlinked nature in the research, while ‘biogeography’ shows clear connections with geographical keywords, illustrating the spatial aspects of the field (Figure 6).

Thematic identification through LDA analysis

Our Latent Dirichlet Allocation (LDA) analysis, detailed in Figure 7, S1 and S2, has delineated five distinct thematic clusters that are representative of the current research landscape in vertebrate palaeontology. These clusters are characterised by a combination of geochronological, anatomical, and phylogenetic terms that align with key research areas in the field:

Figure 6. Keywords cluster network in vertebrate palaeontology from 2013 to 2022. Each node represents a keyword, with lines indicating the connections between them. The thickness of the lines denotes the strength of the association, while the colours correspond to different clusters. Keywords are categorised into 13 clusters, with the core of each cluster representing the main interconnected keywords within that field.

Figure 7. Results of the LDA analysis displaying the five primary topics and their most relevant terms, summarising the thematic findings.

1. Cretaceous Period Dinosauria and Palaeobiogeography: Emphasised by terms such as ‘new’, ‘formation’, ‘cretaceous’, ‘species’, and ‘taxon’, this cluster underscores an intensive investigation into the Cretaceous period’s faunal diversity and paleobiogeographic patterns. The presence of these terms suggests a rich vein of research focused on the systematics and evolutionary history of dinosaurs, shedding light on their ecological niches and distribution during this geological epoch.

2. Tertiary Mammalian Phylogeny and Dental Morphology: The recurrent terms ‘species’, ‘new’, ‘miocene’, ‘genus’, along with specialised dental terminology point towards a robust body of work dedicated to the phylogenetic reconstruction of Tertiary mammals. This theme reflects a detailed focus on dental morphology as a tool for discerning evolutionary relationships and adaptive strategies within this group.

3. Avian and Early Vertebrate Morphology: Characterised by terms like ‘morphology’, ‘evolution’, ‘birds’, and ‘skull’, this thematic group indicates a scholarly focus on the morphological evolution of birds and and early vertebrates throughout the geological timeline. This cluster is particularly revealing of the integrative approaches, a combination of different study methods, combining morphological with phylogenetic analyses to understand the developmental and evolutionary intricacies of these taxa.

4. Broad-Scope Vertebrate Development and Osteology: With terms such as ‘size’, ‘body’, ‘bone’, ‘growth’, and ‘fossils’, this theme delves into a comprehensive study of vertebrate growth patterns and bone structure that transcends stratigraphical periods and taxonomic classifications, encompassing a wide array of vertebrate taxa. It speaks to the palaeontological efforts to reconstruct life histories through the study of growth rings, bone microstructure, and size variation, providing insights into the life processes of extinct species.

5. Pleistocene Palaeoecology and Early Human Occupation: Dominated by ‘pleistocene’, ‘fossil’, ‘north’, ‘america’, and chronological markers, this cluster accentuates the interdisciplinary research at the intersection of palaeoecology and archaeology. It highlights the Pleistocene epoch as a critical period for understanding the dispersal and adaptation patterns of early humans, with a notable focus on the Americas, and their interactions with contemporaneous faunas.

The convergence of these thematic areas, as evidenced by LDA model, demonstrates the multidisciplinary convergence in contemporary vertebrate palaeontology, showcasing the field’s evolution from traditional descriptions to encompassing cutting-edge methods like synchrotron imaging and developmental biology. It is however noteworthy, that the themes identified by LDA are probabilistic and not definitive, and they represent one of many possible interpretations of the complex references in this field. As such, the thematic clusters we present here should be viewed as a guide rather than a conclusive delineation of the field’s research foci. This thematic synthesis not only encapsulates the historical scope of palaeontological research but also underscores the evolutionary narratives that are central to the field. The identified clusters resonate with the profound shifts in palaeontological research focus, from classic descriptive taxonomy towards a more holistic understanding of vertebrate life through time.

Discussion

Advancements and challenges in vertebrate paleontology

Although the Palaeobiology Database (PBDB) has been successful as a broad palaeontological database, it has limitations in terms of data on vertebrate palaeontology (e.g. missing data, incorrect classification; Sallan et al. 2018; Pan et al. 2023). A glaring gap existed in the form of a lack of an open, research-oriented vertebrate palaeontology database on the internet. To fill this void, the DeepBone database project started in 2018 (Pan and Zhu 2019). Unlike PBDB, which has notable gaps in vertebrate palaeontology data, DeepBone focuses specifically on vertebrate palaeontology data. Every entry is meticulously curated by experts and students specialising in vertebrate palaeontology, operating with a specimen-centred approach, backed by referenced references, and framed within a scored systematic tree – making it a uniquely specialised database for vertebrate palaeontology. This endeavour required a considerable human resource investment to extract specimen and taxonomic unit-related data from formally published references. This has set the stage for advancing data science in vertebrate palaeontology, a theme further explored in this study. The establishment of DeepBone represents a proactive stance towards improving the quality and accessibility of data in the realm of vertebrate palaeontology. Based on the accumulated volume of data entries from previous years, the DeepBone database now encompasses nearly all formally published vertebrate palaeontology references from the past decade. This breadth and depth of references make it particularly suited for conducting this bibliometric study.

The yearly volume of articles underlines the discipline’s development, showing an overall upward trend in vertebrate palaeontology (Bornmann and Mutz 2015). Consistent with the exponential growth in publication numbers observed across various academic disciplines over the past century, this trend aligns with broader academic patterns (Fortunato et al. 2018). The increase in the number of articles published, which in 2022 was three times that of 2013, might also reflect broader changes in the academic publishing ecosystem, such as the increase of publishing outlets and heightened competition among publishers and researchers. This suggests that while the attention and participation of scholars in vertebrate palaeontology have been continuously increasing, signalling a stage of rapid development, factors such as the proliferation of scientific journals, competition among publishing companies, and the pressures of ‘publish or perish’ on academic careers may also play a role (Rawat and Meena 2014; Altbach and De Wit 2019). This trend might also be attributed to the continuous progress of science and technology, the constant expansion of the research field, and the continuous increase in major discoveries (Gao et al. 2017; Coccia 2020). However, the number of publications in certain years, especially from 2020 to 2022, has shown a downward trend, likely impacted by the COVID-19 pandemic (Riccaboni et al. 2022).

Fishes, as the earliest vertebrates (Carroll 1988), hold significance evolutionary importance, a trend echoed in the LDA analysis, which also highlighted a focus on Late Mesozoic dinosaurs.

Global landscape and interdisciplinary evolution

Understanding the co-occurrence frequency of references across different countries and institutions offers valuable insights into the global landscape of vertebrate palaeontology research. Our study indicates that substantial contributions come from regions with extensive fossil resources and from countries with strong research infrastructure. However, it’s important to note that contributions to palaeontology are not confined to any specific economic or geographic categories. The scope and impact of research can be influenced by a variety of factors, including the size of the country and its geological resources. Furthermore, the dynamics of international collaborations in palaeontology are complex and multifaceted, as pointed out by Cisneros et al. (2022) and Raja et al. (2022). These collaborations can vary greatly in nature, reflecting a range of inter-country relationships in the scientific community.

The United States of America and the European Union are central in the research collaboration network for vertebrate palaeontology, highlighting the importance of strong international partnerships. The visualisation also reflects China’s growing engagement with these major research contributors. This network includes countries representing large surface areas and both single and multiple political entities. Collaborations in Europe, for instance, often involve multiple countries working together, as exemplified by the European Research Council (ERC)’s roles in fostering joint efforts across various nations, enhancing the research and publication potential. While, recent shifts in global dynamics, such as the UK’s changing relationship with the EU, may reconfigure collaboration patterns, the emphasis remains on the need for adaptability and strong ties between countries and regions are key for progress in vertebrate palaeontology research. Notably, despite Brexit, UK researchers continue to participate and collaborate in major European funding programmes (Naujokaitytė, 2021; Royal Society 2023).

Emerging trends and future directions

Publications in vertebrate palaeontology are found in both specialised and general journals, indicating that the field is evolving beyond its traditional boundaries and is increasingly engaging in multidisciplinary collaborations. While the discipline of vertebrate palaeontology has been gaining interdisciplinary interest, the relatively lower impact factors of major journals in this field suggest that there’s still potential for greater academic recognition. The impact factor of major journals in vertebrate palaeontology is generally not high, compared to those related to some now routine technics (such as synchrotron micro-tomography, the journals in physics having a higher IF than palaeontology publications). However, the specialised journals within vertebrate palaeontology tend to have a higher Cited Half-life than comprehensive journals. This suggests that articles within these journals maintain their academic relevance over a longer period, indicating a lasting impact on the field. This also suggests that assessing the academic value of journals should go beyond impact factors alone, incorporating a variety of metrics.

Classic research in vertebrate palaeontology has traditionally centred on the description of new taxa and the analysis of fossil morphological structures, with a focus on unravelling their evolutionary trajectories. However, recent trends suggest not so much a decline, but a shift in how morphological studies are presented, with many now incorporated into Supplementary Information, while the main text addresses overarching evolutionary problems, especially in journals with limited page numbers like Science and Nature. This change reflects the field’s evolution towards a multifaceted and integrated approach (Sallan et al. 2018). Utilising Latent Dirichlet Allocation (LDA), we further explored key thematic areas in vertebrate palaeontology. While these themes, such as ‘Late Mesozoic dinosaurs and palaeontology’, ‘Cenozoic mammals’ evolution and classification’, ‘anatomy and evolution of birds and early vertebrates’, ‘growth patterns in ancient vertebrates’, and ‘Pleistocene and early human history’, have been foundational in vertebrate palaeontology since its inception in the early 19th century, they continue to be vibrant areas of research today. These enduring topics are being re-examined and expanded with the incorporation of new technologies and interdisciplinary approaches, showcasing the evolving nature of the field. This analysis not only highlights the diverse interests in vertebrate palaeontology but also provides insights into the current directions and evolving trends of the field. Keyword frequency analysis in vertebrate palaeontology provides direct insights into prevalent terms, but its results may be biased due to overused terminology and the subjectivity of researchers in keyword selection. For instance, terms like ‘evolution’ often skew results due to their common usage. Additionally, the lack of standardised keyword selection introduces potential biases. In contrast, LDA investigates deeper, identifying topics based on word co-occurrence patterns. However, interpreting LDA requires a strong professional background, as evidenced by its ability to highlight specialised topics like ‘Late Mesozoic Dinosaurs’ and ‘Tertiary Mammals’ Evolution’. In essence, while keyword frequency analysis identifies general trends, LDA offers a more detailed thematic understanding, contingent on a robust knowledge base.

Over the years, vertebrate palaeontology has integrated diverse methodologies – including morphology, microtomography, genetics, and development – that enhance the field’s comprehensive understanding of its subject matter (Fan et al. 2020; Andreev et al. 2022; Hamm et al. 2022; Kjær et al. 2022). Recent refinement of integrative approaches has deepened vertebrate palaeontological research. The field of vertebrate palaeontology has evolved towards a multifaceted and integrated approach, embracing advanced technologies. For example, three-dimensional reconstruction and computerised tomographic scanning enable precise, non-destructive analysis of fossil ultrastructure (e.g. Gai et al. 2011; Dupret et al. 2014; Andreev et al. 2022). Sedimentary DNA analysis has emerged as a key tool, circumventing the unpredictable preservation of vertebrate fossils and broadening research scope (Kjær et al. 2022). Additionally, big data and deep learning technologies are becoming increasingly valuable for exploring macroevolutionary patterns (Fan et al. 2020; Yu et al. 2022). These advancements underscore the field’s commitment to elevating research quality and academic rigour.

The shifting focus of vertebrate palaeontological research over the last decade provides insight into the field’s evolution. In the period from 2013 to 2016, our data indicates an emphasis on the taxonomy and morphology of ancient organisms, as evidenced by the frequent usage of keywords such as ‘phylogeny’, ‘anatomy’, and ‘skull’ (Figure 4). This reflects a modern phase in the field where these classical aspects of vertebrate palaeontology continued to be a major area of study. However, it is important to acknowledge that, as discussed earlier, detailed morphological analyses have increasingly been relocated to Supplementary Information sections in some journals, indicating a shift in publication practices. From 2017–2019, rising mentions of ‘development’ and ‘integration’ reflect an increased adoption of integrative approaches incorporating genetics, developmental biology, and digital techniques. More recently, the emergence of terms like ‘growth’ and ‘bird’ in our bibliometric analysis suggests a growing interest in specific aspects of vertebrate life, potentially reflecting new research directions in areas such as avian physiology and biomechanics. Tracing such changes through bibliometric data reveals vertebrate palaeontology’s expanding scope and integration with new theories and tools. Additional targeted studies of different eras could provide an enriched understanding of the forces and factors propelling the field’s progression.

The patterns and trends identified in this analysis can serve as a roadmap for scholars in vertebrate palaeontology. Recognising areas of increasing interest, such as the growing focus on certain taxonomic groups or emerging interdisciplinary studies, equips researchers to align their efforts with the evolving landscape of the field. Furthermore, by highlighting potential areas of research that are underrepresented, we provide insights that can guide funding decisions, collaboration opportunities, and research endeavours, aiming to fill these gaps in the coming years. This also implicitly reveals ‘dark spots’ that may become new research niches in the future.

In our bibliometric analysis, a noteworthy pattern emerged regarding the representation of various taxa in vertebrate palaeontology, particularly highlighting a disparity in the fossil records of birds and amphibians compared to reptiles. This disparity is not merely a reflection of biological diversity but also underscores the intricate interplay between an organism’s habitat, physical characteristics, and the likelihood of fossilisation. Birds, for instance, predominantly terrestrial and aerial, face a lower probability of preservation due to their fragile skeletal structure and environments less conducive to fossilisation. Amphibians, despite their aquatic habitats, share a similar fate due to their generally small size and delicate bones.

Moreover, the findings raise intriguing questions about the taxonomic categorisation of transitional forms in palaeontological research. The decision to classify forms like Tiktaalik as fish rather than amphibians, or feathered dinosaurs as reptiles instead of birds, can significantly skew our vision of bibliometric evolution for each arbitrary taxonomic group. This classification bias suggests a potential underrepresentation in studies and publications, which could be rectified by re-examining and possibly reclassifying these transitional forms.

The analysis also sheds light on the geographical and political dimensions of vertebrate palaeontological research. Larger countries, with expansive sedimentary deposits, inherently possess a greater potential for palaeontological discoveries. However, this is further complicated by factors such as colonial history and international collaborations, which have historically influenced the distribution and accessibility of fossil specimens. This geopolitical aspect suggests that the visibility of the fossil record, as we know it, although primarily a product of biological and geological processes, is also influenced by historical and political circumstances (Cisneros et al. 2022; Raja et al. 2022).

While this bibliometric analysis provides quantitative insights into the research landscape, expert interpretation of the results remains crucial, as statistics alone cannot capture the nuances and context of scientific advancement. A balanced approach integrating quantitative data and qualitative assessment may better illuminate the forces shaping vertebrate palaeontology.

Conclusions

Our bibliometric analysis (2013–2022) reveals an increase in vertebrate palaeontology research activity and publications, highlighting the evolution of vertebrate palaeontology from basic descriptive studies to more sophisticated, integrated approaches that integrate various data types. This progression demonstrates the field’s dynamic nature and its shift towards a multifaceted and integrated approach. The LDA analysis, delineating five thematic groups, underscores this shift, reflecting the field’s broadening spectrum of research focus, particularly on reptiles, mammals, and fish. Significantly, our findings reveal the emergence of new research areas like Biodiversity Conservation, Biochemistry, and Molecular Biology, marking a pivotal expansion towards diverse scientific domains within vertebrate palaeontology. This expansion is supported by key contributions from entities with rich resources and robust infrastructures, such as China, Argentina, the United States, and the United Kingdom, the European Union, which have been instrumental in fostering the field’s growth and diversity. Despite facing challenges from the lower impact factors of major journals, vertebrate palaeontology exhibits remarkable resilience and adaptability, maintaining its scientific integrity and contribution to knowledge. This resilience is particularly evident in how the field navigates the complex relationship between research quality and journal prestige, emphasising the importance of content over metrics. However, it’s important to note that our analysis, while comprehensive, is primarily centred on publication trends and thematic developments. The limitations stemming from the available data and the scope of our bibliometric methods restrict a deeper exploration into the underlying causes of these trends and the nuances of global collaboration.

Availability of data and material

All relevant data that support the findings of this study are available from the Corresponding author (Zhaohui Pan, [email protected]), upon reasonable request.

Acknowledgments

We would like to thank Aijin Yu, Shicheng Luo, Tianqin Wang and Fuqi Nie from Qujing Normal University for their assistance in the data verification, and to Min Zhu, Wenjin Zhao, and Per Ahlberg for their invaluable feedback on our initial draft.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research is funded by the National Natural Science Foundation of China [92255301, 42072026, 42130209, 42002015], the Youth Innovation Promotion Association CAS (2021070), the Basic Research Program of Yunnan [202201AU070017], Yunnan Provincial Department of Education Science Research Fund Project [2023J1039], Joint Special Project for Basic Research of Local Undergraduate Universities in Yunnan Province [202101BA070001-209, 202101BA070001-279], The Joint Special Project of Science and Technology Innovation between Science and Technology Bureau in Qujing and Qujing Normal University [KJLH2022YB03], Yunnan Province workstation of Meemann Chang, Academician of the Chinese Academy of Sciences [202205AF150002], and China Scholarship Council [2020004910252]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.