The process of corpus collection consisted of four steps: i.

querying online databases;

First, we defined the query to obtain water resources research about Latin American and Caribbean (LAC) countries. We selected a peer-reviewed literature database based on the following criteria: (1) inclusion of journals from LAC (e.g. SciELO); (2) number of query results in English, Spanish, and Portuguese; and (3) expert assessment of the results with a focus on relevancy. Web of Science and Scopus databases were chosen. Results from the query were assembled into .ref data files.

Second, in accordance with institutional licensing, EndNote was used to retrieve documents from the assembled reference files. This method of corpus collection showed an unequal return rate across the three languages (Table S1 in the Supplement), which was corrected for in the subsequent steps.

Third, we performed a quality assessment of the corpus to assess possible bias between query results and document retrieval within distinct sources or time periods (Year). We defined the within-corpus bias as the difference in relative frequencies f between the query and corpus, such that 1B(Year)=Δfquery(Year)−fcorpus(Year),2B(Source)=Δfquery(Source)−fcorpus(Source). A corpus with satisfactory quality would be one presenting minimal bias from Eqs. (1)–(2). To address the possibility that our automated approach could result in between-corpus bias in terms of return rate, we augmented the corpus by manually downloading articles to correct for biases by year and presence of digital object identifier (DOI). We defined the English corpus as well-behaved and used it as a reference to adjust the Spanish and Portuguese corpora so that each language's corpus reached similar return rates. This adjustment reduced both within- and between-corpus bias by using systematic sampling with unequal inclusion probabilities of the pi , where pi was defined using the normalized product of the bias functions and i refers to the ith non-retrieved article. This procedure was performed twice leading to a similar return rate across languages (Table S1).

Lastly, we performed a cleaning process to prepare the article texts for the topic model. Texts were converted into a standardized version in which all cases were lowered, words not found in a dictionary were removed, and special patterns for words such as emails and URLs were assigned tags. Additional cleaning removed punctuation (except for apostrophes and hyphens inside words) and single-letter words. Finally, lemmatization (the reduction of a word to a common base form) was performed using TreeTagger software by changing all nouns into singular form and all verbs into the infinite tense . This final version of each corpus, with lemmatization, was then used as input for topic modeling.

To build a database of socio-hydrologic country descriptors, we collected 43 relevant indicators from five indexes to compare and contrast LAC countries (Table S2, Data S1). We selected the indicators from the following databases, chosen for their international recognition and global or LAC-wide breadth of data:

AQUASTAT, (http://www.fao.org/nr/water/aquastat/tables/index.stm, last access: May 2021);

Results from the literature review were ground-truthed with an electronic survey sent to corresponding authors of the articles from the corpus. The survey aimed to shed light on researchers' characteristics by including questions about their research experience, institutional affiliation, publication history, their perceptions on funding and interdisciplinarity in the field of water resources, and open-ended questions. The survey received exemption from the University of California, Davis, IRB Administration (ID 1335782-1).

The survey was designed to include 16 questions, in three languages (English, Spanish, and Portuguese), and take approximately 5 min to complete (Table S3). Respondents were asked questions about their position; institutional affiliation; years of experience; main research discipline; countries of birth, residence, and research focus; number of peer-reviewed publications; motivations for picking journals for publications; source of funding; and opinions regarding interdisciplinary research.

The survey was sent in November 2018 to corresponding authors from 22 324 papers, a subset of the final corpus, using Qualtrics survey distribution software. This included articles written in each language: 20 332 English, 1293 Spanish, and 699 Portuguese. The survey was resent to non-respondents weekly until February 2019, leading to 1969 responses. The survey response data was cleaned and prepared for analysis. The survey responses in Spanish and Portuguese were translated to English and compiled into one document.

We clustered countries based on socio-hydrologic variables using two different methods: k-means clustering and hierarchical clustering . We performed the clustering with Euclidean distances and following Ward's criterion. We then investigated the optimal number of clusters by evaluating the evolution with the number of clusters of the total within sums of squares and of the average silhouette width . In addition, we used the following four validation metrics to assess the stability of the clustering under the complete set of clustering variables and performed an iterative procedure where one variable is removed from the set, an approach akin to leave-one-out cross-validation:

The average proportion (APN) measures the proportion of observations not placed in the same cluster under both cases and evaluates how robust the clusters are under cross-validation .

Each document in the corpus was augmented by generating and mining metadata. The mined metadata correspond to author keywords as well as the citing and cited literature resulting in a citation network. The generated metadata correspond to modeled topics and study location.

This section details assessments of the following: i.

ground truth from analyzing survey data;

LAC countries were clustered based on socio-hydrological characteristics using hierarchical and k-mean clustering. Both clustering methods yielded similar results. The total within sums of squares evolved after two clusters were chosen. Similarly, the average silhouette width strongly exhibited a peak for two clusters. Further inspection of clustering in principal component dimensions indicated that the cluster with Mexico and Brazil was significantly distinct from all other countries, explaining the observation of a sharp peak in average silhouette width. However, validation metrics exhibited optimal null values of average proportion (AP) and average distance between means (ADM) for two or three clusters. In addition, average distance (AD) and figure of merit (FOM) were lower for three clusters than for two. Based on these results, we chose three clusters to describe the grouping of countries based on their socio-hydrologic variables.

The most common categories of answers in the comment section were summarized (Table S7). Difficulties related to funding stood out as the main challenge for respondents (Fig. ). These were often linked to political or economic trends in their countries. Respondents further explained that available funds are not only distributed unequally across countries but also within countries and across research areas. Collaboration, or the lack thereof, was another important theme. Many respondents claimed that they collaborated more with North America, Europe, and Australia than with other LAC countries, in large part due to funding opportunities. Many indicated an interest in collaborating with researchers in other LAC countries.

First, we look at research volume by country and topic. The cumulative research growth showed an exponential growth trend over the last 4 decades, although the trend was more evident in cluster 1 (Brazil and Mexico) than in clusters 2 or 3 (Fig. ). These results were used to interpret periods of time in which the scientific production of water research was increasing or decreasing on a region-wide scale, in reference to the growth trajectories of each socio-hydrologic cluster. Reasons for this growth could relate to a wide array of changes in LAC over the last 40 years beyond an increasing interest in addressing water-related knowledge gaps: increased access to the internet and online resources, availability of existing research in online repositories, and expansion of opportunities in higher education .

A residual analysis identifies three distinct periods since 2000, the first year that each socio-hydrologic cluster was represented by over 30 research papers in each language. Annual output was lower than each cluster's general trend for the first few years, followed by a period of relatively higher output from 2007–2012, ending with a trend of decreasing growth from 2013–2017, although some residuals are below a single standard deviation within these periods. It is possible that uniform anomalies below or above general growth trends correspond to region-wide events, although a several-year lag could be possible between causal events and effects in research output. For example, a connection may exist between Brazil's economic crisis starting in 2012 and the subsequent drop in research output from 2013–2017.

Combining research topics with predicted study location describes the composition of water research in LAC with a chord diagram (Fig. ). This chord diagram was obtained from the weighted bipartite network between countries and topics (Fig. ). For legibility, the network was filtered to remove the edges with weight lower than the 75th percentile of the edges of a given country. In other words, for each country, only the top 25 % links are displayed.

While research about Brazil, Mexico, Argentina, and Chile are bright spots that dominate the research landscape, the absence of countries in the Caribbean and most of Central America indicates a shortage of research in these regions. A country's socio-hydrologic cluster correlates to its representation in overall research, with cluster 1 (Brazil and Mexico) receiving the most research, followed by cluster 2, then cluster 3, which includes most of Central America and the Caribbean. Although population size likely affects each country's representation in the overall research output, it does not precisely correlate with research volume. We therefore expect that other factors used to define the country clusters (e.g. a country's water and economic resources, geography, and history) influence the likelihood that researchers study that country.

Mexico and Argentina alternate for second highest representation, depending on the topic, after Brazil. These findings are confirmed by survey respondents, of which 35 % study Brazil, followed by Mexico with 15 %, and Argentina with 9 % (across all research topics). Only 12 % of survey respondents focused most of their research on countries in the Caribbean or Central America.

Water research is not distributed equally among disciplines and is primarily conducted in the physical and life sciences, representing together 80 % of topic probabilities. Survey responses also confirm these findings, as 80 % of respondents identified as life and physical scientists. The lack of water research in social sciences may reflect a combination of low publication rates (compared to physical sciences), disciplinary preference for publishing in books rather than peer-reviewed articles, a historical framing of water management as a purely technical discipline , and financial resources from governments biased towards physical sciences and engineering.

Next, we look at the spread of water research in LAC, described by topic normality for general, specific, water budget, and method topics. Importantly, Caribbean nations and most of Central America did not have enough research to be included in these analyses and represent blind spots. For general topics (Fig. a), physical and life sciences represent bright spots, with topics spread normally across both countries and documents. Conversely, underlining findings from research volume (Fig. ), the limited spread of social sciences across documents represents a blind spot. For specific topics (Fig. b), hydrology, atmospheric sciences, statistics, and mathematics are bright spots across countries and documents, while marine biology and forest science are likely blind spots. The majority of specific topics have high normality across countries (>0.8) but low normality across documents (<0.4). These topics are either infrequently mentioned or, if mentioned, are the sole subject of a paper, lacking integration in interdisciplinary research. For water budget topics (Fig. a), precipitation, which must be monitored and understood to manage water resources, has a distribution closest to normal, representing a bright spot, while glaciers is far from having a normal distribution, likely because few countries have glaciers to study (i.e. Argentina, Bolivia, Chile), and they are relatively less relevant for other countries' water budgets. For method topics (Fig. b), statistics, quantitative methods, and water sampling are identified as bright spots of water research methods across the region.

The least normality is seen in two topics of great importance for water management: reservoirs and risk assessment. Both topics have normality values far below 1 across both countries and documents, suggesting poor representation of these topics on a broad scale (Fig. ). These are alarming blind spots given the importance of reservoirs for water supply reliability and their impacts on local communities and ecosystems. Furthermore, a lack of research in reservoirs and risk assessment has troubling future consequences as climate change increasingly acts as a risk multiplier, including for reservoir operation. Although it is possible that information on reservoirs and risk assessments exists in gray literature, such as government reports and university publications, this information is not as accessible as scientific publications and limits regional knowledge sharing. These findings confirm similar results in a country-specific analysis of water research opportunities in Brazil .

We complete our description of LAC's water research portfolio by estimating degrees of connectivity. Four network graphs were produced. The network graphs showed a density value of 1; as expected, all topics and countries are connected. The country citation network identified 23 nodes and 529 edges. The network degree of connectivity has a maximum degree of <50 %, medium degree of <30 %, and minimum degree of less than <10 %, where each node corresponds to a LAC country and where colors represent socio-hydrological clusters. The general topics network identified 5 nodes and 25 edges, with a degree of connectivity from a maximum of 45 % to a minimum of 27 %. The specific topics network has 43 nodes and 1849 edges, with a maximum degree of connectivity of 18 %, a medium degree of 10 % and a minimum of 5 %. The water budget and topics network has 27 nodes and 729 edges, and a maximum degree of connectivity of 5 %.

The high volume of research about Brazil (45 % of the labeled English corpus) motivated further investigation to see if this large scientific output is proportionally more influential than the research about other countries. Brazil is a central bright spot of the citation network (Fig. ), as citations by publications from all countries, including Brazil itself, are primarily directed toward research about Brazil (49 %). Furthermore, the geographic location or socio-hydrologic cluster of a country do not appear to indicate the extent of citations of researchers working in that country. For example, although Argentina and Chile share a particularly long border and a rich history of political and economic ties, the citation network reveals only a medium degree (14 %) of connectivity between the two countries. Similarly, countries in the Caribbean and Central American regions that share many similarities, including water management challenges, show a low degree of connectivity (less than 5 %).

Results from our survey complement findings from the citation network. Over half of participating researchers collaborated with researchers outside of the country(ies) they study. Despite being the only Portuguese-speaking country, Brazil was the country most often listed in collaborations within LAC (17 %). Brazil's prominence may be partially explained by the legacy of relationships formed during graduate-level training when Brazilian researchers study abroad and when graduate students from other South American countries study in Brazil. Despite its greater connectivity, the review of 250 water science papers presented at the 2019 Brazilian Water Resources Symposium still found a lack of a common scientific agenda within the country and a need for more interdisciplinary research and collaboration with international communities, “especially with other Latin American countries with shared water issues” .

A quarter of research collaborations involve non-LAC researchers, mostly in the United States (14 %). This may reflect differences in access to funds and highlights how more affluent countries can influence the scope of research conducted in LAC. Respondents indicated that insufficient and precarious funding arrangements are their main challenge. 89 % of respondents said that the government is their main source of funding, which may explain that a country's political and economic context were mentioned as further aggravating funding availability. Funding difficulties were also associated with a lack of value given to water research and to the long timeframes associated with research that are misaligned with decision-makers' timelines.

Furthermore, being physically close or part of the same socio-hydrologic clusters did not increase the likelihood of cross-country collaboration between countries. For instance, 22 % of researchers in the neighboring countries of Argentina or Chile either study both countries or collaborate with one another, while 24 % of those researchers report collaboration with Brazil. Researchers from Mexico and Brazil, who share a socio-hydrologic cluster, collaborate even less, with only 14 % reporting to work or collaborate in both countries, despite a high level of connectivity from Mexico to Brazil in the citation network. Conversely, more than 80 % of researchers in the Caribbean reported collaborating with researchers from other Caribbean nations and few collaborated with Brazil. This is in opposition to the findings from the citation network, showing few citations within the Caribbean and more frequent citations of publications on Brazil, but this could partially be due to the limited number of articles studying the Caribbean region included in our corpus.

We assessed the connectivity of water research throughout the region by aggregating research from all countries by topic (Fig. ). Separate networks are presented to illustrate connectivity between the five general topics (Fig. a) and specific topics (Fig. b). A reciprocal relationship between the physical and life sciences dominates the citation network, while research within social sciences, engineering, and mathematics is cited less often (Fig. a). Research connectivity from all subdisciplines is heavily polarized towards research on hydrology and water resources; however, certain topics display a comparable degree of self-citation (Fig. b). This may indicate that water resources research communities in political science and governance, civil engineering, ecology, geophysics, and toxicology are siloed or more connected to non-water resources research communities.

Interestingly, behind these few silos and the central node, a vast network of connectivity exists. While this level of connectivity is low (less than 10 %), it supports the characterization of water resources as a scientific discipline where research topics are already integrated , albeit with room for strengthened interdisciplinarity. Survey respondents confirm this low level of background connectivity, reporting dissatisfaction with the interdisciplinarity of their research. Nonetheless, an opportunity exists to build off these existing connections to form communities of researchers and strengthen research impact through knowledge sharing and collaboration across disciplines .

The wide scope of this study, intended to capture the breadth of the state of water resources research across LAC, required inevitable compromises in the depth of information and the subsequent ability to thoroughly interpret our results. Notably, much scientific literature in Spanish and Portuguese was not readily available or accessible online, and this resulted in the need to rely on English publications as a proxy of research across LAC. A targeted method to collect gray literature would increase the size of the Spanish and Portuguese corpora. Of the literature we found, very little focused on Caribbean countries, and this lack of information limited subsequent analysis. A targeted method of corpus augmentation and human-read validation towards less-represented countries and topics will likely increase the model's predictive capabilities and may improve the representation of Caribbean countries.

In addition, while the presented citation network included all LAC countries, the exclusion of countries outside of LAC prevented a more comprehensive analysis of LAC countries' reliance on non-regional research. Survey responses suggested that reliance on non-LAC research was high, as researchers stated they were more likely to collaborate with scientists outside of LAC than within LAC. Inclusion of non-LAC countries in the analysis of scientific interactions could present many opportunities for expanding our findings in future research. Finally, our study indicated where bright and blind spots appear across research in LAC, but did not aim to examine causal relations for these patterns, a common shortcoming in Science of Science . Survey results indicate that funding for research is an important driver influencing where bright and blind spots occur.

However, a more comprehensive answer would require exploring historical, political, economic, and social dynamics influencing the allocation of research resources. Overall, this work displays the value of our novel method to interpret results from machine learning, points to the need for a deeper and wider understanding of existing water resources research in water vulnerable regions, and warrants expanding our methods to include gray literature and coverage across the Global South.