May 17, 2023

Antoinette Abeyta¹, Jason McLachlan², Jody Peters², Nicholas R. Record³, Anna R. Sjodin^4§, Olivia Tabares⁵, Alyssa M. Willson²
Co-authors are listed alphabetically since all contributed substantially to this project.

¹University of New Mexico, Gallup, ²University of Notre Dame, ³Tandy Center for Ocean Forecasting, Bigelow Laboratory for Ocean Sciences, ⁴Environmental Protection Agency (EPA), ⁵Universidad Nacional Autónoma de México
^§These views are my own and do not reflect the opinions or beliefs of the EPA.

SUMMARY

Here, we introduce a way to evaluate barriers to inclusion in ecological forecasting and environmental sciences using the iterative forecasting and adaptive management cycle, and suggest ways to extend our understanding of ecological forecasting beyond this cycle. We begin by highlighting three examples of barriers to inclusivity (i.e., hypotheses, models, and language). Next, in an attempt to make ecological forecasting a more inclusive discipline through the very ways we conceptualize its component steps, we reimagine the iterative forecasting cycle to emphasize and center marginalized groups. Finally, we provide suggestions for next steps that focus on working with students and reversing marginalization of historically excluded individuals. 

We invite anyone with an interest in participating to join these efforts. If you have comments or suggestions or would like to participate as a co-author in a manuscript that builds from these ideas, reach out to us in the comments below, at info@ecoforecast.org, on the #inclusion channel on the EFI Slack group, through Twitter (@eco4cast) or directly contact any of the authors of the blog post.

INTRODUCTION

Ecology, one of the pillars of ecological forecasting, is fraught with experiences of racism and sexism, despite concerted efforts to improve inclusion and reduce barriers to entry (Martínez-Blancas et al., 2023). One of the tasks from the DEI Strategic Plan (developed by the Ecological Forecasting Initiative’s (EFI’s) Diversity, Equity, and Inclusion (DEI) Working Group) is to identify barriers hindering historically excluded individuals’ participation in ecological forecasting and other quantitative environmental sciences. The EFI DEI working group has been discussing this in monthly meetings, at the EFI annual meetings in 2021 and 2022, and at national meetings (e.g., Geoscience Alliance and the Ecological Society of America meetings in 2022). Our goal in this blog post is to spark perhaps new considerations for the EFI community about what the barriers to ecological forecasting are and ways to begin addressing those barriers.  We want to open up a discussion with the broader EFI community, and those interested in ecological forecasting, and ask for input on the barriers and the next steps identified, and ask for input about barriers and solutions beyond those that we have already considered.

Iterative ecological forecasting in the tradition promoted by EFI is often conceptualized as a cycle that has focused on the technical requirements or research outcomes of forecasting (e.g., Dietze et al., 2018) (shared here in Figure 1). We used this conceptual diagram of the iterative forecasting and adaptive management cycle as a starting point for thinking about the human components required for forecasting and to identify where the cycle presented barriers to entry and persistence in ecological forecasting and in what ways.  ​​Often existing forecasting cycles center research outcomes, leaving community as an afterthought. To make ecological forecasting more reflective of the communities we work with, we have to structure our forecasting cycles to center values that are important to the communities we serve. 

For example, at the Geoscience Alliance meeting in 2022, we spent considerable time discussing how marginalized communities often do not have access to the internet, computer hardware, and cyberinfrastructure tools that are often taken for granted in the ecological forecasting community (Federal Communications Commission, 2012), which corresponds to part B of the iterative forecasting and adaptive management cycle. The list of barriers to persistence in ecological forecasting quickly became too long for a blog post, as we realized that the forecasting cycle reflects the systemic barriers to inclusivity that persist in ecological forecasting and quantitative environmental sciences more generally. This led to the question: Can we re-envision a forecasting cycle that incorporates inclusion throughout the forecasting enterprise? 

THE BARRIERS

Hypotheses – Who gets to decide what gets forecasted, and how?

Hypotheses, as they are imagined in the Western scientific process, originate within a specific epistemology, or way of knowing and understanding. However, there are many epistemologies, and without engagement with communities, forecasts developed from and motivated by historical ecological literature will be biased towards what has historically been studied and the tried and true approaches to scientific inquiry. In ecology as in many scientific disciplines, the historical literature is dominated by White men. Additionally, the topics and studies that have produced the most amount of data and positive results (Nakagawa et al. 2022), or the data streams that provide consistently collected, archived, and available data, will be the most represented and continually reinforced. An alternative is co-produced forecasts, which center the assumptions, knowledge, and ideas most valuable to a community’s needs (Nyadzi et al., 2022; Record et al., 2022).

Funding similarly reinforces perpetual barriers to entry for historically marginalized communities. Research funds that continually reward novelty and the bleeding edge of science often don’t align with local or community needs (Flagg, 2022; Van Horne et al., 2023), and there is often no incentive to continue to work with local communities on persistent problems after the grant cycle ends. Moreover, researchers from historically excluded groups, often more in touch with local or community needs or working in the Global South, get their work published and showcased less than scientists in Global North institutions (Smith et al., 2023), thus perpetuating barriers to publishing and funding research related to community problems.

Models – Who generates the model, and who has access to the models and data?

Models are, by definition, simplifications of the truth. The aspects of truth that are simplified are inherently biased by those generating the models, and models are often developed to reflect a researcher’s hypothesis about how a system works. The more aspects that must be omitted to create a working model, the more opportunities exist for removing perspectives and truths understood by marginalized people. Generation of models is subject to the same epistemological weaknesses as hypotheses because models in many ways are our hypotheses. For ecological systems, specifically, mechanisms are complex, and inclusion of epistemological diversity becomes increasingly important for contextual understanding in such complex systems (Page, 2014). Importantly, mechanistic understanding and decisions based on information produced in partnership among epistemologies are more robust (Berkes, 2009; Schuttenberg and Guth, 2015; Wheeler and Root-Bernstein, 2020). So, in addition to the inherent value of a more inclusive process (Morrison and Steltzer, 2021; NASEM, 2022), incorporating different perspectives can provide a more holistic, and arguably stronger, understanding of natural systems. 

Additionally, creation of the models that produce forecasts is influenced by who has access to use or run the models. Not all models are open source and many have a historical legacy of ownership which can make it more difficult for individuals new to the field to learn how to run a certain model. Even models that can be built from scratch (e.g., linear regression models) require statistical knowledge that is often inaccessible to scholars outside of research intensive academic settings.

Data ownership can also be a barrier to creating forecasting models. On one hand, there are many open source data sets (e.g., NEON). However, these data may be too big to access on personal laptops, the data are often not collected equally across locations (e.g., NEON), and the data that have been collected will be biased towards the interests of those who have set up the collection processes. There are also cases where data sovereignty needs to be considered (Vera et al., 2019) and where data should be kept proprietary (e.g., working with culturally relevant data or human subjects). Similarly, running models and making probabilistic forecasts requires access to computational resources and expertise. Basic quantitative training is not universal nor created equal (Willson et al., 2023).   

Language

El uso del inglés como lingua franca en el contexto científico, impone barreras a quienes no son hablantes nativos del idioma en diferentes aspectos. Desde la dificultad para adquirir nuevos conocimientos en una lengua diferente a la materna, pasando por limitación de oportunidades de estudio y laborales, hasta sesgos de publicación  y experiencias de discriminación en instituciones académicas (Woolston & Osorio 2019). 

A lo largo del siglo XX se ha observado que el inglés pasó de ser parte de un modelo plurilingüe de comunicación científica internacional, en el que otros idiomas, como el francés y alemán tambien eran comúnmente utilizados, a tener un dominio casi absoluto en la comunicación de las ciencias naturales y sociales, lo que limita cómo y quienes pueden comunicar ciencia e incluso aprenderla (Hamer, 2013; Amano et al 2021).

En el caso particular del Ecological Forecasting, al aprendizaje de tópicos complejos como la estadística, modelación matemática y programación, se añade la dificultad de hacerlo en inglés para hablantes ESL (English as a Second Language), ya que la mayoría de los textos, artículos y recursos de apoyo (e.g. foros, videos, tutoriales) están en este idioma (Amano et al. 2021). Prueba de ello es el repositorio que como Education Section de EFI hemos colectado, con materiales exclusivamente en inglés. Si bien esto está sesgado por la composición lingüistica de nuestro grupo, es un reflejo de muchos tópicos en ciencias.

Si bien la enseñanza del idioma inglés suele estar en la curricula de la gran mayoría de países no angloparlantes desde la educación básica, y un 20% de la población global se encuentra estudiando inglés como segunda lengua (TEFL Academy 2020), la comprensión necesaria del idioma para poder aprender y producir predicciones ecologicas dificilmente puede ser garantizada para personas que no hayan recibido instrucción privada y vengan de contextos socioeconómicos provilegiados. Esto impone dificultades a las personas interesadas en investigación que no hablen inglés tanto para adquirir nuevos conocimientos como para hacer posgrados, colaboraciones de investigación y publicar en revistas internacionales (Woolston & Osorio 2019; Amano et al 2021).

Nota: Proporcionaremos una traducción al español de toda la publicación tan pronto como los coautores tengan tiempo de crear la traducción.

Note: Following our own recommendations, we will provide a Spanish translation for the entire post as soon as the co-authors have time to create the translation.

MAKING INCLUSION EXPLICIT IN FORECASTING CYCLES

The forecasting cycle shown in Figure 1 has appeared in different iterations in subsequent studies (e.g. Moore et al. 2022). We’ve added two alternative forecasting “cycles”, described below, to this list, with a goal of reducing barriers to inclusivity. As we considered alternative figures we asked, “What should it look like? Should it be a cycle at all, or some other conceptual model?” The two figures presented here should be considered as first drafts and provide some food for thought on barriers to inclusivity in the forecasting process and how to address them. Moving forward, we hope that the redrawing of the ecological forecasting framework could, itself, be an inclusive process. We present figures depicting forecasting cycles in this slide deck, where we welcome additional contributions from readers and invite anyone interested to join our efforts to write a manuscript associated with this project.

Community-Centered Forecasting Cycle – Example 1

Science is often lauded for its objectivity, but each hypothesis or question carries with it inherent biases that reflect the values, thought processes, and experiences of the researcher(s), as discussed above in The Barriers section. These values show when scientific findings are applied, as in forecast-based decision-making, and the end user is often different from the researcher, with different values, thought processes, and experiences. In the DEI working group at EFI, we think a lot about these differences and how to unite them for the common goals of improving the field of ecological forecasting and promoting informed ecological decision-making. We recognize not only the inherent value of inclusivity (Morrison and Steltzer, 2021), but also the benefit that different experiences and expertise can have on an emerging field like ours (Woelmer et al. 2021, Willson et al., 2023).¹ 

With inclusion front of mind, we created figures such as Figure 2 to focus the forecasting cycle within a diverse community of practice. Figure 2 stresses centering project goals within the values of the researchers and end users (blue circle). Ethics, therefore, cannot be an afterthought: the continuous evaluation and re-evaluation of goals, values, and ethics should be a discrete action item. These biases are then explicitly acknowledged as informing the scientific process (the outer red circle), and vice-versa (purple arrows). Importantly, those involved in the forecasting research effort can join the process from either the blue or ourter red circle, demonstrating the interdisciplinary qualities of forecasting. And finally, decisions can be made without science, but they cannot be made independently of personal values and biases (black internal arrows).

¹ The value of diversity and inclusivity goes far beyond just ecological forecasting. Also, see the DEI statement on the EFI DEI Working Group webpage.

Community-Centered Forecasting Cycle – Example 2, model of practice at the Tandy Center for Ocean Forecasting

The Tandy Center for Ocean Forecasting works on developing forecasts for communities, industries, and other users. In thinking about inclusivity, individuals at the Tandy Center are thinking about those communities that are affected by the forecasts they create and the decisions based on them. In the conventional forecasting cycle (Figure 1), scientists are centered. However, some of the lessons that ocean ecosystem forecasters have learned and have been generous to share over the years have had to do with unintended consequences and accidental harms caused by well-intentioned forecasting programs (Hobday et al., 2019). Being a forecaster can be precarious, and it’s probably impossible to avoid every pitfall, but hopefully, we can learn from some of these lessons. The Center works under the belief that being more inclusive can help align the needs of communities using, or affected by, the forecast with the design of the forecasting program. The schematic included below (Figure 3) comes from the Tandy Center’s guidance documentation (Record, 2022). It’s not exactly a cycle, but more like a map to guide dialogues with different community groups so that they can collaborate at each step. There’s also a video that walks through the figure. In short, communities should be partners throughout the process, contributing to the design of the system. By centering forecast users rather than forecasters, we hope to help forecast development be more accessible and reach new people and places that might be under-resourced or otherwise excluded from the mainstream of forecasting applications.

Notable weaknesses that still remain

Of course, including everything in a single figure is challenging (hence this blog post). Continued discourse about how to address ongoing weaknesses is shared in Next Steps. Additionally, salient ideas have been left out of or de-emphasized in the above figures but have been discussed throughout the EFI DEI working group meetings. For example, diversity, equity, inclusion, and justice (DEIJ) efforts are often listed in grant applications as Broader Impacts but are not considered to contribute to intellectual merit. A more thorough list, including some more inclusive alternatives, can be found in Table 1.

NEXT STEPS

Thinking about students

The legacy of barriers to inclusivity has been carried on through science education, and breaking this legacy needs to include thinking about education. Today’s students are highly motivated, eager, and ready to tackle broadly defined diversity, equity, inclusion and justice issues. This passion is shaped from a range of experiences and identities. Good education and pedagogical approaches are attentive to the diversity of learners (Harris et al., 2020; Miriti, 2019; Rawlings-Goss et al., 2018). Teaching takes many forms, but can consist of (1) active and collaborative learning activities (Corwin et al., 2018; Graham et al., 2013), (2) providing student agency and voice (they are co-creators in results), (3) honoring existing knowledge, (4) avoiding deficit language when teaching. Additional resources for inclusive teaching can be found in the Inclusive Pedagogy Resources compiled by EFI as well as this extensive list of inclusive teaching resources and strategies from the University of Michigan’s Center for Research on Learning and Teaching.  There are also efforts within the EFI community and the EFI Education Working Group to compile open educational resources (Willson and Peters, 2021; Willson et al., 2023, Table 1), as well as to develop educational modules in collaboration with faculty at Minority Serving Institutions to teach data science tools that incorporate Traditional Ecological Knowledge and cultural values. 

Reversing marginalization

Data science continues to have tremendously low rates of representation from historically excluded and marginalized groups. Consequently, data and computational tools are often created from a narrow world view of priorities, values, and practices (see The Barriers section for more details). When these tools are used on marginalized communities, they are often limited to the interpretations, biases, and preconceived notions of the creators of these tools (David-Chavez and Gavin, 2018). Without centering the knowledge, experience, and perspectives of marginalized groups in the creation of these tools, they become tools of oppression, promoting erasure, perpetuating stereotypes, and continuing violence and harm to communities. If we want ecological forecasting to become a tool to enact meaningful and just change, we have to structurally alter our research methods and practices through a lens of intersectionality to center the voices of marginalized communities, make them leaders in the creation of tools, and coproduce models and tools with communities (Crenshaw, 2014). In Table 1, we demonstrate how structural modifications to research practices can improve engagement with marginalized groups, showing the shift of power towards community benefit.  By encouraging the forecasting community to thoughtfully consider how to develop collaborations, we hope future research will center and bring in the perspectives of historically excluded individuals.

Connect with us!

Developing this post has been a learning process for the entire EFI DEI Working Group and we acknowledge that this is an ongoing and iterative process where it is important to hear from additional voices and to continue to learn.  We hope that this post can help the scientific and EFI community continue to think about barriers to participating in forecasting and environmental sciences and solutions for overcoming those barriers. We welcome comments, suggestions, and feedback on these ideas presented. Our goal is to turn this post into a manuscript that builds from these ideas and we invite anyone that would like to share comments or participate as a co-author in the manuscript effort to reach out to us in the comments below, at info@ecoforecast.org, on the #inclusion channel on the EFI Slack group, or through Twitter (@eco4cast).

Table 1. We provide a list of different topics related to forecasting (and data science and science in general) that demonstrate how structural modifications to research practices can improve the inclusion of historically marginalized groups. The three columns represent situations where there is the most room for improvement in connecting to and centering marginalized groups, where there is room for improvement, and a column with situations that are most beneficial to the most people. It is important to note that this table is written as a generalization, and all research practices in a project should be decided collaboratively with the community and be aligned with their values. This table was inspired by The Wheel of Power and Privilege and other related work (Hierarchy of Indigenous Data, the Global Indigenous Data Alliance, and models for decolonizing science research, e.g., David-Chavez 2019, David-Chavez et al. 2020). You can also view a PDF of the Table HERE.

Topic	Situations with the Most Room for Improvement	Situations with Room for Improvement	Situations that are the Most Beneficial and Works to Center and Bring People in
Community Involvement	– Research is done on communities without input or with limited input from the community – Research dollars are not directed to the community – Researchers are outsiders to the community	– Coproduction of knowledge between researchers and the community – Research dollars are distributed to work with the community – Researchers include community members – Addresses generational and long-scale impacts	– Marginalized communities are leading research initiatives – Research dollars go directly toward marginalized groups – Community members are PIs on research projects – Research involves children and younger generations in the development and execution of projects
Education	– Eurocentric education practices – Value is placed on credentialed programming only – Programs limited to R1 institutions and primarily white institutions – High cost of tuition is inaccessible – Materials are often presented in English only	– Education emphasizes different cultures and values – Improved access to higher education – Programs available to public, community, and tribal institutions.  – Inclusion of bilingual materials	– Curriculum material is culturally informed and relevant – Recognition of the impact of non-credentialed programs – Improving access to credentialed/advanced degree programs – Education programs span the spectrum of education (K-12 through post-graduate programs) to promote generational learning – Provides education outside of academic institutions – Traditional ecological knowledge incorporated into education – Materials are created and translated into multiple languages used by the community
Benefits & Harms	– Using tools or data that misrepresent a community – Science is done on others without consent – Perpetuating violence, harm, and erasure of marginalized groups – Communities and people are an afterthought in projects – Values are centered on Western ideals – Does not acknowledge the communities where research comes from	– Science is done with the consent of the community but without input in the design – Research provides knowledge for the community – Prioritizes and centers most privileged communities or most represented communities – Land acknowledgments recognize the legacy of colonization	– Working on science in tandem with communities to benefit communities – Giving agency to marginalized communities to define and access research – Centers people and communities in projects – Values are centered on communities ideals – Land back, or land acknowledgments address the continued harm and benefit to the institution
Computation & Technology	– Requiring subscriptions or software  – Limited training options – Limited computational and internet access	– Using low-cost and accessible tools – Providing access to computers or mobile devices – Use of universal design in materials	– Resources developed for individuals without internet access in mind – Open-source materials are used or generated – Computational resources are readily accessible
Data Availability (including journals & tools)	– Data is only available through private access or behind a paywall – Communities and individuals are unable to control data or access to data – Requires specialized software, tools, knowledge of where to access data – Data collected without consent of community or individuals	– Data available via request rather than open online access – Data and tools are published for sharing for further research and collaborations – Data is collected with community knowledge and consent	– Promotes ethical use of data – Acknowledges and supports data sovereignty – Broad education on how to access tools, resources, and data – Data collected with respect to cultural values and practices – Improves the ease of accessing public data – Data is made open upon the wishes and needs of the community
Diversity & Justice	– Homogenous racial or cultural research teams – Research perpetuates harm to marginalized groups or maintains the status quo – Funding is focused on short-term impacts – Projects ends when funding ends	– Researchers work with students from diverse backgrounds – Improved representation of marginalized groups in research spaces	– Diverse teams with agency – Research enacts meaningful social change – Funding and research acknowledge the importance of long-term impacts – Project implementation continues after funding ends

Citations

Amano, T., Rios Rojas, C., & Boum, I. I. Y, Calvo M, Misra BB. 2021. Ten tips for overcoming language barriers in science. Nature Human Behaviour, 5(9), 1119-1122.

Berkes, F., 2009. Indigenous ways of knowing and the study of environmental change. J. R. Soc. N. Z. 39, 151–156. https://doi.org/10.1080/03014220909510568

Corwin, L.A., Prunuske, A., Seidel, S.B., 2018. Scientific presenting: Using evidence-based classroom practices to deliver effective conference presentations. CBE Life Sci. Educ. 17, es1. https://doi.org/10.1187/cbe.17-07-0146

Crenshaw, K. 2014. On Intersectionality: Essential writings. The New Press. 

David-Chavez, D.M., Gavin, M.C., 2018. A global assessment of Indigenous community engagement in climate research. Environ. Res. Lett. 13, 123005. https://doi.org/10.1088/1748-9326/aaf300

David-Chavez DM. 2019. A guiding model for decolonizing environmental science research and restoring relational accountability with Indigenous communities (Doctoral dissertation, Colorado State University).

David-Chavez DM, Valdez S, Estevez JB, Meléndez Martínez C, Garcia Jr AA, Josephs K, Troncoso A. Community-based (rooted) research for regeneration: understanding benefits, barriers, and resources for Indigenous education and research. AlterNative: An International Journal of Indigenous Peoples. 2020 Sep;16(3):220-32.

Dietze, M.C., Fox, A., Beck-Johnson, L.M., Betancourt, J.L., Hooten, M.B., Jarnevich, C.S., Keitt, T.H., Kenney, M.A., Laney, C.M., Larsen, L.G., Loescher, H.W., Lunch, C.K., Pijanowski, B.C., Randerson, J.T., Read, E.K., Tredennick, A.T., Vargas, R., Weathers, K.C., White, E.P., 2018. Iterative near-term ecological forecasting: Needs, opportunities, and challenges. Proc. Natl. Acad. Sci. 115, 1424–1432. https://doi.org/10.1073/pnas.1710231115

Federal Communications Commission, 2012. Eighth Broadband Progress Report (No. FCC 12-90).

Flagg, M., 2022. Reward research for being useful — not just flashy. Nature 610, 9–9. https://doi.org/10.1038/d41586-022-03131-7

Graham, M.J., Frederick, J., Byars-Winston, A., Hunter, A.-B., Handelsman, J., 2013. Increasing Persistence of College Students in STEM. Science 341, 1455–1456. https://doi.org/10.1126/science.1240487

Hamel, R. E. 2013. El campo de las ciencias y la educación superior entre el monopolio del inglés y el plurilingüismo: elementos para una política del lenguaje en América Latina. Trabalhos em linguística aplicada, 52, 321-384. 

Harris, B.N., McCarthy, P.C., Wright, A.M., Schutz, H., Boersma, K.S., Shepherd, S.L., Manning, L.A., Malisch, J.L., Ellington, R.M., 2020. From panic to pedagogy: Using online active learning to promote inclusive instruction in ecology and evolutionary biology courses and beyond. Ecol. Evol. 10, 12581–12612. https://doi.org/10.1002/ece3.6915

Hobday, A.J., Hartog, J.R., Manderson, J.P., Mills, K.E., Oliver, M.J., Pershing, A.J., Siedlecki, S., 2019. Ethical considerations and unanticipated consequences associated with ecological forecasting for marine resources. ICES J. Mar. Sci. 76, 1244–1256. https://doi.org/10.1093/icesjms/fsy210

Martínez-Blancas, A., Bender, A., Zepeda, V., McGuire, R., Tabares, O., Amarasekare, P., Mastretta-Yanes, A., Miriti, M., Santos, A.M.C., Vaz, M.C., 2023. Surviving Racism and Sexism in Academia: Sharing Experiences, Insights, and Perspectives. Bull. Ecol. Soc. Am. 104, e02033. https://doi.org/10.1002/bes2.2033

Miriti, M.N., 2019. Nature in the eye of the beholder: A case study for cultural humility as a strategy to broaden participation in STEM. Educ. Sci. 9, 291. https://doi.org/10.3390/educsci9040291

Moore TN, Thomas RQ, Woelmer WM, Carey CC. Integrating Ecological Forecasting into Undergraduate Ecology Curricula with an R Shiny Application-Based Teaching Module. Forecasting. 2022 Jun 30;4(3):604-33.

Morrison, D.L., Steltzer, H., 2021. Diverse values, philosophies and ideas beget innovation and resilience in ecology and for our world. Ecol. Appl. 31, e02351. https://doi.org/10.1002/eap.2351

National Academies of Sciences, Engineering, and Medicine (NASEM), 2022. Next Generation Earth Systems Science at the National Science Foundation. National Academies Press, Washington, D.C. https://doi.org/10.17226/26042

Nakagawa, S., Lagisz, M., Jennions, M. D., Koricheva, J., Noble, D. W. A., Parker, T. H., Sánchez-Tójar, A., Yang, Y., & O’Dea, R. E. 2022. Methods for testing publication bias in ecological and evolutionary meta-analyses. Methods Ecol. Evol. 13, 4– 21. https://doi.org/10.1111/2041-210X.13724

Nyadzi, E., Werners, S.E., Biesbroek, R., Ludwig, F., 2022. Towards weather and climate services that integrate indigenous and scientific forecasts to improve forecast reliability and acceptability in Ghana. Environ. Dev. 42, 100698. https://doi.org/10.1016/j.envdev.2021.100698

Page, S.E., 2014. Where diversity comes from and why it matters? Eur. J. Soc. Psychol. 44, 267–279. https://doi.org/10.1002/ejsp.2016

Rawlings-Goss, R., Cassel, L. (Boots), Cragin, M., Cramer, C., Dingle, A., Friday-Stroud, S., Herron, A., Horton, N., Inniss, T.R., Jordan, K., Ordóñez, P., Rudis, M., Rwebangira, R., Schmitt, K., Smith, D., Stephens, S., 2018. Keeping data science broad: Negotiating the digital and data divide among higher education institutions. Workshop Bridg. Digit. Data Divide.

Record, N.R., Evanilla, J., Kanwit, K., Burnell, C., Cartisano, C., Lewis, B.J., MacLeod, J., Tupper, B., Miller, D.W., Tracy, A.T. and White, C., 2022. Benefits and Challenges of a Stakeholder-Driven Shellfish Toxicity Forecast in Coastal Maine. Frontiers in Marine Science, p.1031.

Record, N., 2022. Early Warning Systems for Harmful Algae: A Stakeholder-Centered Framework. https://doi.org/10.13140/RG.2.2.24501.14568

Schuttenberg, H.Z., Guth, H.K., 2015. Seeking our shared wisdom: a framework for understanding knowledge coproduction and coproductive capacities. Ecol. Soc. 20(1). http://www.jstor.org/stable/26269755

Smith, O.M., Davis, K.L., Pizza, R.B., Waterman, R., Dobson, K.C., Foster, B., Jarvey, J.C., Jones, L.N., Leuenberger, W., Nourn, N., Conway, E.E., Fiser, C.M., Hansen, Z.A., Hristova, A., Mack, C., Saunders, A.N., Utley, O.J., Young, M.L., Davis, C.L., 2023. Peer review perpetuates barriers for historically excluded groups. Nat. Ecol. Evol. 7, 512–523. https://doi.org/10.1038/s41559-023-01999-w

TEFL. 2022. English language teaching statistics. Retrieved from https://www.theteflacademy.com/blog/english-language-teaching-statistics/

Van Horne, Y.O., Alcala, C.S., Peltier, R.E., Quintana, P.J.E., Seto, E., Gonzales, M., Johnston, J.E., Montoya, L.D., Quirós-Alcalá, L., Beamer, P.I., 2023. An applied environmental justice framework for exposure science. J. Expo. Sci. Environ. Epidemiol. 33, 1–11. https://doi.org/10.1038/s41370-022-00422-z

Vera, L.A., Walker, D., Murphy, M., Mansfield, B., Siad, L.M., Ogden, J., 2019. When data justice and environmental justice meet: formulating a response to extractive logic through environmental data justice. Inf. Commun. Soc. 22, 1012–1028. https://doi.org/10.1080/1369118X.2019.1596293

Wheeler, H.C., Root-Bernstein, M., 2020. Informing decision-making with Indigenous and local knowledge and science. J. Appl. Ecol. 57, 1634–1643. https://doi.org/10.1111/1365-2664.13734

Willson, A., Peters, J. Ecological Forecasting Repository of Online Educational Resources. https://doi.org/10.25334/XAW7-JW16

Willson, A.M., Gallo, H., Peters, J.A., Abeyta, A., Bueno Watts, N., Carey, C.C., Moore, T.N., Smies, G., Thomas, R.Q., Woelmer, W.M., McLachlan, J.S., 2023. Assessing opportunities and inequities in undergraduate ecological forecasting education. Ecol. Evol. 13, e10001. https://doi.org/10.1002/ece3.10001

Woolston, C., Osório, J. (2019). When English is not your mother tongue. Nature, 570(7760), 265-267.

Woelmer W.M., L.M. Bradley, L.T. Haber, D.H. Klinges, A.S.L. Lewis, E.J. Mohr, C.L. Torrens, K.I. Wheeler, A.M. Willson. 2021. Ten simple rules for training yourself in an emerging field. PLOS Computational Biology 17(10): e1009440. https://doi.org/10.1371/journal.pcbi.1009440