A Bibliometric Analysis of Forest Gap Research during 1980–2021

Abstract

(1) Background: Forest gaps play an important role in promoting forest regeneration and facilitating the forest growth cycle. Since the 1980s, forest gaps have been widely studied by forestry scientists. The purpose of this study was to review the global literature from 1980 to 2021, based on the scientific database Web of Science Core Collection, and to summarize the research hotspots and the trends of the forest gaps. (2) Method: A bibliometric analysis was performed using the visual analytic software CiteSpace to quantify the description of annual publications, collaboration analysis of authors, institutions and countries, co-citation analysis of cited journals, cited authors, and cited references. The keyword co-occurrence, burst, and time zone were also analyzed by the software. (3) Results: The results show that the volume of annual publications is increasing. Dr. Harald Bugmann is the author with the most published works. The most active institution is the American Forest Service. The United States, Canada, and China are the three most productive countries. “Ecology” is the most cited journal. The results indicate that the hotspot in the forest gap research has shifted, and the effects of forest environmental changes caused by forest gaps under climate change have received more attention from scientists. In the future, more attention may be paid to the role of forest gaps on near-natural forest management patterns, the effect of forest gaps on forest sustainable development, and the way to study forest gaps using lidar technology. (4) Conclusion: Our results can help to understand emerging trends in forest gap research to inform forest ecology and management.

1. Introduction

Forest gaps are openings in the forest canopy caused by the death of trees due to disturbances [1]. Gaps are usually small disturbance events and are considered the critical stage of forest regeneration and succession according to the “forest growth cycle theory” [2]. The forest cycle theory considers forest communities as a mosaic of patches at various stages of development in space and dynamically changing in time, such as seasonal changes and mosaic renewal [3]. Forest gaps produce brighter and warmer habitats due to a significant increase in irradiance [4]. When gaps appear, the composition of different light-demanding regeneration species is affected. The regeneration patterns of woody species in the expanded gaps indicate that shade-tolerant species dominate in the low-light areas (small gaps), and shade-intolerant species dominate in the high-light areas (large gaps) [5]. Gaps accelerate species replacement, change forest structure, and increase the functional diversity of the ecosystem [6]. Compared with woody species in the understory, those in gaps have higher growth and survival rates [7]. Moreover, gaps may affect the microbial community and enzyme activity of soil [8]. Based on the characterization of individual functional groups of microorganisms, small gaps have moderate soil biogenicity, while large gaps have weak soil biogenicity [9].

Forest gaps have received widespread attention from forestry, ecology, and botany scholars. A growing number of international scholars have conducted studies on forest gaps [4,10,11]. Some high-impact reviews contribute to in-depth research in specific directions, including gap measurement, gap regeneration, and biodiversity in gaps [12,13,14,15,16]. However, a comprehensive review of the historical development of forest gaps is lacking. Therefore, a bibliometric analysis is necessary to review the study of forest gaps. The importance of bibliometric analysis through data mining, information processing and visualization techniques to show current advances, emerging trends, and research frontiers is increased compared to traditional reviews and theoretical reviews [17]. Bibliometric analysis improves the accuracy of scientific literature and avoids the bias of research by adding objectivity and legitimacy to literature reviews [18].

This study aimed to track research trends in forest gaps based on bibliometric analysis with the visualization software CiteSpace. The development of quantitative bibliometrics allows for better visualization. The following research questions guided the research process: (1) Identify core authors, institutions, countries, and journals; key citations; high-frequency and burst keywords. (2) What are the trends in finding research topics related to the forest gap by keywords?

2. Materials and Methods

2.1. Data Collection

We searched the core database using the Web of Science from the SCI-E and SSCI databases on 20 April 2022. The search strategy was as follows: The topic was “forest gap*” OR “canopy gap*” OR “treefall gap*” OR “gap dynamic*” [2]; the time span was from 1 January 1980 to 31 December 2021. A total of 3567 publications were initially obtained. Subsequently, the writing language was English, and the document type was set to “Article” or “Review,” which included 3469 publications. Then, irrelevant categories were removed, and secondary screening was performed by browsing the title of the literature to manually eliminate irrelevant documents. A total of 3363 articles were screened, including 3261 research articles and 102 reviews. We extracted the metadata of these articles, including the title, keyword, abstract, information of author, journal, citation, and institutional affiliation for further analysis.

2.2. Data Processing

CiteSpace is a scientific visualization package based on a programmed Java application [19]. The software can detect the development of scientific literature, especially those caused by triggering events [20]. In addition, it provides metrics of structural and temporal properties of networks, clusters, and nodes based on information from the literature. Structural metrics include centrality, modularity, and silhouette index. The modularity and silhouette measurement are used to describe the properties of the overall structural network [21]. The effectiveness of knowledge maps can be evaluated based on modularity and silhouette values [17]. Modularity indicates a reasonable clustering division structure, and modularity (Q) > 0.3 indicates a significant division result. The silhouette reflects the degree of data cluster. The clustering result is reasonable: when silhouette (S) > 0.5, the data is highly clustered with silhouette close to 1; with silhouette close to −1, the data is not reasonably clustered; and with silhouette close to 0, the data is on the boundary of two natural clusters [20,21]. The centrality index measures how close a node is to two or more other nodes in a network. Temporal metrics include burstiness, which indicates a sudden surge of interest in the research community at a particular node during a particular period [22,23]. Burst words are keywords that appear suddenly within a certain period, and greater strength of the burst means greater popularity.

The software version was 5.8. R3 was used for the analysis of dynamic and multidimensional networks. The respective knowledge maps were drawn. Node types in CiteSpace were set to “author”, “institution”, “country”, “cited journal”, “keyword” and “cited reference”. The thickness of nodes linked to each other indicates the strength of co-occurrence, collaboration or co-citation [24]. Different colors and sizes of tree rings represent different years and numbers of publications (citations). Moreover, cluster analysis on keywords and cited references was performed to identify research hotspots and frontiers, with areas representing different years. A cluster of classes was named after the core keyword node, and the number decreased as the members increased. The time slice was set from 1980 to 2021, and the year of each slice was used in this study [25]. Depending on the type of analysis, different node types were selected. Selection criteria were defaulted as g-index = 25, Top N = 50, and Top N% = 10%. The threshold parameters included citation threshold (c), co-citation threshold (cc), and cosine coefficient threshold (ccv), which were set to 2/2/20, 4/3/20, and 4/3/20, respectively. Citation threshold (c) denotes the minimum number of citations or frequency of occurrence, co-citation threshold (cc) denotes the frequency of co-linear or secondary citations within a given time slice, and cosine coefficient threshold (ccv) denotes the co-linear rate or co-leading rate. CiteSpace provided many visualization options, including straightforward interface options and interpretation. Finally, many studies were converted into visual maps by CiteSpace for better analysis, and Excel 2016 was applied to depict the annual publications.

3. Results

3.1. Annual Variations in Publications

The trends and development status can be reflected by changes in the number of publications annually in a specific study area [20]. On the basis of annual publications and the slopes of the fitted growth models (Figure 1), the number of publications can be divided into three research stages (the initial, rapid, and deep development stages). The initial development stage was 1980–1987, with an average annual publication of 2.25 and a slope of 0.405. The definition of forest gaps was focused on this stage [26,27]. The rapid development stage was 1988–1998, when the average annual publications increased to 52.63 with a slope of 9.527. During this period, the characteristics and regeneration of the gap received increasing attention [12,28]. The deep development stage was 1999–2021, with the average annual publications increasing to 120 and a slope of 3.134, suggesting a lower growth rate than the rapid development stage. In the deep development stage, new methods and theories were applied to forest gap research (UAV and forest dynamics theory driven by small-scale disturbances) [29,30,31,32].

3.2. Collaboration Analysis

3.2.1. Analysis of Authors

The analysis of core authors’ distribution may supply a better overview and promote academic research advancement and collaboration [33]. The larger the nodal circle in the network, the more the number of articles published by the author. The close collaboration between authors can be determined by the density of the lines. Dr. Harald Bugmann (30) is the author with the most articles in the searched database, indicating that he is the most active scholar in the field of forest gaps, followed by Dr. Shin-Ichi Yamamoto (24) and Dr. Christian Messier (21). Moreover, the color of the nodal circles also shows that Dr. Bugmann is the author with many publications in recent years. About 20 of his articles are based on forest gap models to predict changes in forest structure under global climate change. Figure 2 shows the authors with more than 10 publications. Not all major collaborative networks are connected, and most scholars in this field show fragmented connections.

3.2.2. Analysis of Institutions

The difficulty for single scholars to have comprehensive expertise and research resources requires the collaboration of research institutions to address increasingly complex scientific problems. Therefore, the analysis of research institutions can provide insight into the core research institutions and enhance their influence through cooperation with core institutions. The United States Forest Service, the Chinese Academy of Sciences, and the University of Wisconsin rank in the top three for publication, indicating the essential role these institutions play in linkages and communication (Figure 3).

Table 1. Top 10 institutions based on the number of documents.

Rank	Number of Documents	Centrality	Year	Institution
1	133	0.22	1998	US Forest Service
2	86	0.13	2003	Chinese Academy of Sciences
3	63	0.04	1997	University of Wisconsin
4	48	0.12	1998	Smithsonian Trop Res Institution
5	39	0.04	1998	University of Helsinki
6	37	0.04	1998	University of Quebec
7	31	0.03	2001	Oregon State University
8	30	0.03	1998	Forestry & Forest Prod Res Institution
9	30	0.03	1998	University of Florida
10	27	0.03	1998	Hokkaido University

also shows the centrality of the top 10 institutions regarding the number of articles published. Centrality is an important measure that can indicate the turnaround in a research field and indirectly demonstrate its impact. Notably, 6 of 10 institutions are universities, suggesting that the primary research forces are distributed in universities. Nevertheless, the centrality is not high, which shows that international cooperation and exchange are relatively weak and further improvements are needed.

3.2.3. Analysis of Countries

Rapidly revealing the impact of countries and their contribution to forest gap research can be obtained by analyzing the number and centrality of publications from different countries [17]. Based on the analysis of country cooperation, a country/territory cooperation network of 161 nodes from 1980 to 2021 was obtained (Figure 4). The United States ranks first with 1327 publications, accounting for 39.4% of all publications, followed by Canada and China with 326 and 233 publications, respectively. These three countries account for 56.4% of the total publications. In addition, among the top 10 countries with the highest number of publications, the United States has the largest centrality, followed by China, indicating their high international status and the importance of their research on forest gaps. However, Canada has published the second most papers among all countries, but its centrality is only 0.07, indicating that it lacks international collaboration despite its achievements in forest gap research (

Table 2. Countries or regions larger than 100 based on the number of documents.

Rank	Number of Publications	Centrality	Year	Country
1	1327	0.58	1982	USA
2	326	0.07	1991	Canada
3	245	0.51	1994	China
4	221	0.07	1985	Japan
5	193	0.08	1995	Germany
6	157	0.02	1996	Brazil
7	149	0.24	1988	England
8	130	0.17	1987	Australia
9	99	0.26	1995	Switzerland
10	98	0.29	1994	France

).

3.3. Co-Citation Analysis

3.3.1. Analysis of Cited References

The document co-citation analysis can help detect the initial, most cited, and influential literature to understand the intellectual basis of each research domain [34]. In this study, an aggregation network consisting of 1906 references and 7551 link lines was constructed, and the top 10 key co-citation clusters were observed in Figure 5 and

Table 3. Top 10 co-citation clusters based on the number of documents.

Cluster ID	Size	Silhouette	Mean (Year)	Representative Terms
0	227	0.881	1995	tropical rain forest;
1	177	0.91	2001	broad-leaved forest;
2	166	0.945	2016	within-gap position;
3	159	0.888	2007	old-growth forest;
4	159	0.92	1990	Seedling ecology;
5	151	0.964	1986	treefall gap;
6	143	0.906	2013	airborne laser;
7	130	0.887	2008	central Amazon forest;
8	83	0.969	1996	forest gap model;
9	74	0.994	1979	beech-sugar maple forest

. Colors correspond to time slices. Each node represents a cited reference. The higher the silhouette score, the better the consistency between cluster members [35].

Clusters #0, #1, #3, #7, and #9 are about the forest types in which gap research is mainly located. Cluster #0 has 227 members with a silhouette value of 0.881. Brokaw et al. found that chance events contribute more to tree species richness than ecological niches in tropical forests [36]. Cluster analysis showed that Cluster #1 has 177 members with a silhouette value of 0.91 from the perspectives of related research on the tropical rainforest. In broad-leaved forests, recruitment opportunities can be neutralized by dense shrub cover in canopy gaps, and diversity can be promoted by those gaps [37]. The references in this cluster are mainly related to the broad-leaved forest. Cluster #3 has 159 members with a silhouette value of 0.888. Prior research has reported that by reviewing the methodology for studying forest gaps, detailed field measurements are recommended to determine the size and age of gaps to provide solutions for silvicultural management in the old-growth forest [10]. The references in this cluster are mainly related to old-growth forests. Cluster #7 has 130 members with a silhouette value of 0.887. The presence of drought and high temperatures in the central Amazon forest increase tree mortality globally [38]. Cluster #9 has 74 members with a silhouette value of 0.994. Previous research explored the growth of species in different-sized forest gaps in the Merced Primeval Forest of eastern North America found that disturbance regimes favored tolerant species in the forests studied [26]. The references in this cluster are mainly related to the beech-sugar maple forest.

Clusters #2, #4, and #5 are the main research objectives. In cluster #2, there are 166 members with a silhouette value of 0.945, which may facilitate the establishment of seedlings by the successive extension of gaps along the sun-exposed gap edge [39]. The references in this cluster are mainly concerned with the within-gap positions. Cluster #4 has 159 members with a silhouette value of 0.92. Prior researchers have suggested that recruitment sites are determined by a combination of tropical forest habitat and survival rates [40]. The likelihood of reaching recruitment sites depends on dispersal patterns, the survival of dormant seeds or juveniles, and the interactions of species with pathogens and predators in forest gaps and surrounding forests. The references in this cluster are mainly related to seedling ecology. Cluster #5 has 151 members with a silhouette value of 0.964. Another researcher examined germination, seedling establishment and growth, and survival of seedlings in treefall gaps with different gap sizes [3]. Then, tree species were divided into pioneer and non-pioneer groups. The references of this study are mainly about the treefall gap.

Clusters #6 and #8 are about the main approach to the study of forest gaps. Cluster #6 has 143 members with a silhouette value of 0.906. The gap data at all heights aboveground from the lidar analysis were used to compare gap-size frequency distributions among forests in the lowland Peruvian Amazon [41]. The references are mainly related to airborne lasers. Cluster #8 has 83 members with a silhouette value of 0.969. Previous research established a simplified forest model applicable to plant population dynamics and soil carbon/nitrogen turnover [42]. This model also provided reliable parameterizations of the abiotic environment. The research suggested that further rigorous model comparison and validation are required to increase the reliability of models. The references of this cluster are mainly about the forest gap model.

Notably, two of the top three cited references are about research methods. The first and second most cited references were published by R Core Team in 2016 and 2019, respectively. The third most cited reference was published by [43]. These researchers verified the contribution of forest gaps to species diversity and investigated the causes of tree species diversity on the Barrow Colorado Island.

3.3.2. Analysis of Cited Authors

Many empirical studies have shown that previous analysis is effective in evaluating the situations of disciplinary development and identifying microstructures of specific fields. It can reveal dynamic changes and future developments [42]. A co-citation takes place when two researchers are cited in the same article. In this study, the CiteSpace-based mapping of co-citations excluded anonymous authors, identifying 537 nodes and 1793 links in the studies about the forest gap (Figure 6).

Influential or leading scholars are those with a high cited rate and high betweenness centrality [44]. Among them, Runkle is the most cited author in the statistics, with 793 citations, but his centrality is only 0.11. Acevedo has the highest centrality of 0.55, with only 2 citations. Higher centrality indicates stronger influence and importance, not high total citations (

Table 4. Top 10 cited authors based on centrality.

Rank	Number of Documents	Centrality	Year	Cited Author
1	2	0.55	1984	Acevedo
2	50	0.26	1980	Bormann
3	44	0.15	1981	Barden
4	143	0.12	1984	Nakashizuka
5	739	0.11	1981	Runkle
6	140	0.11	1980	Watt
7	3	0.11	1980	Anderson
8	654	0.1	1982	Denslow
9	305	0.1	1981	Lorimer
10	49	0.1	1982	Hartshorn

). Runkle defined two types of forest gap (canopy gap and extended gap) [26,45]. The results showed that saplings, size distribution, and species composition prior to gap formation can help determine which species will dominate the gap. In addition, high rates of repeated disturbance benefit the species growing at moderate light levels and surviving several periods of suppression before reaching the canopy [46,47]. Acevedo designed a transition model to better simulate the dynamics of landscape and species-rich forests [48]. The model could be derived from the gap model by defining states based on species, functional roles, vertical structure, or other convenient cover types.

3.3.3. Analysis of Cited Journals

When articles are cited more frequently, the academic value of the journal is higher [49]. In this study, 322 journals were obtained, and 402 journals with the citation relationship were identified using the citation knowledge map (Figure 7). The journals with high citation rates are Ecology, Forest Ecology and Management, and the Journal of Ecology. Potentially revolutionary scientific publications were identified by high centrality values and are highlighted by purple rings in the software CiteSpace [50]. Journals with high centrality are Ecology, Ecological Monographs, and the Canadian Journal of Forest Research. Ecology has the most published articles (2588) and the highest centrality (0.73). It belongs to the broad category of Environmental Sciences and Ecology in the Chinese Science Academy system. This result indicated the importance of the journal. The cited articles from Ecology focused on forest community dynamics. The analysis of the journals showed that most articles in the field of forest gaps were included in the category of environmental sciences and ecology (

Table 5. Top 10 cited journals based on the number of documents.

Number of Documents	Centrality	Year	Cited Journal	Major Disciplines
2588	0.73	1981	Ecology	Environmental Science and Ecology
2147	0.08	1989	Forest Ecology and Management	Agriculture and forestry science
2060	0.24	1980	Journal of Ecology	Environmental Science and Ecology
1802	0.47	1987	Canadian Journal of Forest Research	Agriculture and forestry science
1508	0.01	1987	Oecologia	Environmental Science and Ecology
1269	0.64	1981	Ecological Monographs	Environmental Science and Ecology
1231	0.03	1992	Journal of Vegetation Science	Environmental Science and Ecology
1101	0.02	1995	Ecological Applications	Environmental Science and Ecology
1065	0.13	1988	Oikos	Environmental Science and Ecology
1056	0.06	1987	Science	Comprehensive journal

).

3.4. Keywords Analysis

The keywords describe the contents of the core and provide the shortest summary of the contents [51]. The keywords analysis can reveal research hotspots and future trends in the forest gap. In this study, a keyword trend analysis was conducted according to the co-occurrence of keywords, the cluster map, and the outbursts of keywords.

According to the analysis of the keyword co-occurrence (Figure 8), six keywords that have been used for a long time with high frequency (more than 500 citations) are “canopy gap,” “dynamics,” “regeneration,” “growth,” “pattern” and “diversity.” These keywords are the major contents of forest gaps (Figure 8).

For further research, the keyword knowledge map was clustered using “Cluster View” in CiteSpace, which divided the keywords into the top 10 clustering groups. CiteSpace has an automatic tagging feature to characterize derived clusters. In particular, it extracts terms from the keywords in each cluster and uses them as tags [52]. The extracted keywords are the main hotspots in the clusters (

Table 6. Cluster groups of the keyword nodes.

Cluster	Size	Silhouette	Year	Top Terms (LLR)
0	24	0.974	1996	climate change; light; growth; photosynthesis; silviculture
1	24	0.953	2001	northern hardwood forest; dynamics; disturbance
2	19	0.904	1999	habitat selection; fire; population; seed size; management
3	18	0.936	1996	pattern; community; morphology; forest
4	17	0.899	2001	tree; shade tolerance; tree architecture; miconia; rain forest
5	16	0.947	2008	coarse woody debris; photogrammetry; stand; natural disturbance regime
6	16	0.841	2000	boreal forest; snowmelt; root: shoot ratio; mutualism; underplanting
7	16	0.874	2006	vegetation; history; spatial heterogeneity; landscape; wind
8	14	0.944	1997	regeneration; tropical forest; mixed forest
9	14	0.911	2003	carbon; alpine forest; soil; foliar litter; litter decomposition

). The largest clusters are #0 and #1, with 24 publications focusing on “climate change” and “northern hardwood forest”. Clusters 0 and 1 are the only two clusters with more than 20 publications. The earliest Clusters 1 and 3 have 18 publications focusing on “pattern.” More specifically, the top 10 clusters are all shown before the average year 2010. The highest profile value is for cluster 0 (0.974). In addition, related representative topics in this cluster cover “habitat selection,” “tree,” “coarse woody debris,” “boreal forest,” “vegetation,” “regeneration,” and “carbon”.

A research hotspot of particular interest to the scientific community is usually a keyword that has increased significantly in frequency over a short period of time. The analysis of keyword bursts is shown in Figure 9, and the keywords in the table are sorted by the year of occurrence. Red blocks indicate the year of occurrence, and their lengths indicate the duration. With the higher intensity of keyword bursts, the degree of occurrence was higher. Burst detection quantifies the focus on research hotspots and emerging trends. The strength of burst keywords indicates the intensity of interest in a particular area [53]. Regarding burst strength, “lidar (17.88)” is the strongest, followed by “Costa Rica (17.38)” and “impact (16.10)”. The keywords that lasted longer than 10 years are “heterogeneity (11 years)” and “climate change (10 years)”. Compared to other keywords, the above keywords receive greater attention and have a greater impact. In the last 5 years (2016–2021), research on the forest gap has been more active in “mortality (13.74),” “conservation (10.5)”, and “decomposition (9.22)”. Among the 25 words that appeared, 19 are after 2000 (the deep development stage), demonstrating the importance of forest gap research.

4. Discussion

Forest gaps are a common form of disturbance in natural forests and the basis of forest management. The results can help understand the progress of forest gap research, including active journals, core collaborations, and bursting research areas over the past 41 years (1980–2021). The number of publications in forest gap research from 1980–2021 gradually increased and can be divided into three phases (Figure 2). In the third phase (1999–2021), the number of publications essentially reached 100 per year, suggesting that ecologists and foresters are enthusiastic about forest gaps through developing relevant theories, research methods and techniques [54,55,56]. Our study showed that most scholars in the field of forest gap research have relatively fragmented connections with each other. Future research requires stronger collaboration to enhance the impact of international cooperation and research innovation through collaboration. The top 10 research institutions in this area are concentrated in the United States (50%), the most published country (Table 1 and Figure 4), probably because wealthier countries can provide financial support to researchers and institutions [50]. The major journals in forest gap research are Ecology, Forest Ecology and Management, the Journal of Ecology, and the Canadian Journal of Forest Research (Table 5), probably because these journals are international and multidisciplinary.

Citation analysis is a method for studying patterns and frameworks of a particular cognitive subject and identifying key players [44]. Cluster analysis identifies clusters that represent relevant areas of research by grouping similar topics together [53]. We reviewed and classified the top 10 clusters, which can provide a reliable reference for non-experts to quickly gain a comprehensive understanding of forest gap studies. The most frequently cited article is by R Core Team (2016) and provides a statistical computational and graphical system. The article constructs a model to make it easier and more accurate to understand the structure and function of forest gaps [57].

Keyword clustering analysis and burst analysis can assess hotspots and present a holistic view of frontier research issues [20]. An analysis of each phase of keyword co-occurrence reveals that during the initial development period (1980–1987), the main concern was on environmental impacts, such as “light (235)”. In the rapid development stage (1988–1998), the main concern was on the description of forest gaps, such as the “canopy gap (994)” and “regeneration (726)”. In the deep development phase (1999–2021), the concern was more on forest management, such as “management (152)” and “old growth (76)”. The analysis of keyword bursts revealed that the top 25 keywords with the strongest citation burst were mostly in the deep development stage. The largest burst of “lidar” was seen during this stage. With the progress of research, new survey technologies were added to forest gap research. Using lidar for forest canopy gap measurement and description has become a trend in the forest gap research field [58,59,60]. The secondary keyword in terms of burst strength is “impact”. Some articles have discussed the impact of forest gaps on the community in gaps [7], and others have discussed the impact of disturbance on gap formation [61]. The third keyword in terms of burst strength is “mortality.” An increasing number of articles focus on the seedling and sapling mortality process in forest gaps [62]. Although the number of articles in the rapid development stage is the highest, its burst is not as high as that in the deep development stage, indicating that forest gaps play an important role in the development of forest structure. Therefore, the study of forest gaps has important theoretical and practical values in sustainable forest management. The study of forest gaps will not decrease in the future.

5. Conclusions

The critical function of forest gaps in forest ecosystems has attracted global attention. Many studies have assessed the forest gap regarding forest regeneration, biodiversity, and silvicultural control. In this study, a comprehensive and clear visual analysis was conducted for articles related to forest gaps in the core journals of the Web of Science. Since 1980, the number of publications has shown a general growth trend. Most scholars in the field exhibit fragmented connections. The United States Forest Service started first and has the largest number of studies. The United States has been the most productive country with the highest centrality. Canada and China started late in studying forest gaps but grew rapidly in later years to become the second and third most productive countries. Then, the most productive authors and the articles with the highest number of citations and co-citations were identified. The most cited author in the statistics is Runkle. The most co-cited article was entitled R: a language and environment for statistical computing. Regarding research categories, journals in environmental sciences and ecology dominate this research area, followed by journals in agricultural and forestry sciences. Hot topics range from “Costa Rica”, “succession”, and “hardwood forest” to “abundance”, “species richness”, and “establishment” to “impact”, “lidar”, and “mortality”. The results show that with the development of research, scholars pay more attention to the effects of forest environmental changes caused by forest gaps under climate change, and the research objects are becoming more refined. In the future, more attention may be paid to the role of forest gaps on near-natural forest management patterns, the effects of forest gaps on forest sustainability, and the way to use lidar technology to study forest gaps.

Although we assessed the most extensive studies available in the Web of Science database, some studies related to the topic are not available in this database. In addition, some countries have databases in their own languages, such as the CNKI database in China [63]. Although multiple keywords were used in this analysis, the search results were filtered by research direction, resulting in missing some of the results. In addition, the search only included papers and reviews and was conducted based on “topics” rather than “all databases,” leading to the omission of some publications. Finally, only studies published in English were retrieved in this study, and some important studies published in other languages may have been omitted. There are other bibliometric tools available, such as VOSviewer software [64], that can be used with multiple research methods for a comprehensive and comparative analysis. In addition, CiteSpace can quickly display complex citation spaces, research status, hot spots and trends. However, it also has some drawbacks. It can fail to be read during the run due to too little data, thus leading to a failed mapping process. Therefore, this analysis cannot fully replace a systematic search. In conclusion, an in-depth study of the specific content of the literature leads to a more objective and comprehensive scientific analysis of the literature.