{"paper_id":"0dfa52c4-1aa5-4be2-9784-ee2690fa61c5","body_text":"Title: Development and assessment of tailored illustrations to enhance community \nunderstandings of genetics topics  \n \nAuthors: Audrey M. Arner1, Tobias C. McCabe1, Amanda Seyler2, Siti Nurani Zamri3, Tan Bee \nTing A/P Tan Boon Huat3, Kar Lye Tam3, Patriciah Kinyua4, Echwa John4, Sospeter Ngoci \nNjeru4,5, Yvonne A.L. Lim3, Michael Gurven6, Colin Nicholas7, Julien Ayroles8, Vivek V. \nVenkataraman2, Thomas S. Kraft9, Ian J. Wallace10, Amanda J. Lea1 \n \nAffiliations:  \n1Department of Biological Sciences, Vanderbilt University, Nashville, TN, USA, 37232 \n2Department of Anthropology and Archaeology, University of Calgary, Calgary, Alberta, Canada \n3Department of Parasitology, Universiti Malaya, Kuala Lumpur, Malaysia 50603 \n4Turkana Health and Genomics Project, Centre for Community Driven Research - Kenya \nMedical Research Institute, Nairobi, Kenya 54840-00200 \n5Center for Community Driven Research, Kenya Medical Research Institute, Nairobi, Kenya \n54840-00200 \n6Department of Anthropology, University of California Santa Barbara, Santa Barbara, USA \n93106 \n7Center for Orang Asli Concerns, Subang Jaya, Malaysia 45790 \n8Department of Integrative Biology, University of California Berkeley, Berkeley, California, USA \n9Department of Anthropology, University of Utah, Salt Lake City, Utah, USA \n10Department of Anthropology, University of New Mexico, Albuquerque, New Mexico, USA \n \nCorrespondence: amanda.j.lea@vanderbilt.edu \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\nAbstract \nObjectives \n Effective communication about genetics concepts is essential for collaborative \nanthropological genetics research. However, communication can be challenging because many \nideas are abstract and may be especially unfamiliar to communities with limited access to formal \neducation. Indeed, there are no widely adopted models for communicating such information, nor \na clear understanding of the social factors that may shape participant engagement. Here, we \nconducted a qualitative and quantitative, community-driven study to understand how illustrations \ncan be useful to support concept sharing with two Indigenous groups—the Orang Asli of \nPeninsular Malaysia and the Turkana of Kenya. \n \nMethods \nWe used a two phase approach to create and evaluate how illustrations can bolster \ncommunication about genetics concepts. First, we created images illustrating answers to \nfrequently asked questions about genetics, iteratively updating the illustrations based on \nparticipant feedback. Second, we conducted 92 interviews to evaluate the finalized illustrations’ \neffectiveness. Finally, we analyzed the interview data using thematic analyses, multivariable \nmodeling, and multiple correspondence analyses to identify patterns in participant \nunderstanding and feedback, including age, sex, market integration, and schooling. \n \nResults \n Participants reported high interest in genetics research (92%) and broadly positive \nperceptions of the illustrations. Familiar, locally-grounded imagery was preferred and associated \nwith greater perceived clarity, while more technical illustrations were more frequently reported as \nconfusing. Quantitative analyses showed strong internal consistency across measures of \nengagement and understanding, with modest variation by degree of market-integration, \nschooling, and sex. \n \nDiscussion \nOur findings demonstrate that community-specific visualizations, co-developed through \niterative feedback, can effectively support engagement with genetics research in participant \ncommunities.  \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\nIntroduction \n \n Integrating genetics and genomics into biological and biomedical research has advanced \nour understanding of human evolution, disease, and phenotypic variation. However, such \ninsights have not been equally spread across populations and have focused primarily on \nindividuals of European ancestry living in high-income countries [1–4]. The importance of \ndiverse sampling is widely recognized as crucial for robust understanding of both evolutionary \nprocesses and the genetic architecture of complex traits and diseases [5,6], and is essential to \nboth address health disparities and maximize the reach of benefits from downstream \ndiscoveries [7]. This understanding has led to recent initiatives expanding genomic research in \nundersampled populations, for example, work from the H3Africa Consortium [8], Uganda \nGenome Resource [9], and BioBank Japan [10]. In contrast to large-scale initiatives, many \nanthropologists have built both health- and evolutionary-focused genomics studies around \nlong-term relationships with individual Indigenous communities to promote communication, trust, \nand transparency [11–13]. Nevertheless, Indigenous populations around the world continue to \nremain some of the most underrepresented groups in any subfield of genomic research [3,4,14]. \n Indigenous populations are underrepresented in genetics research due to a complex \ninterplay of factors, including historical practices of extraction and exploitation among both the \nbiomedical and anthropological fields, with particularly high visibility cases in the United States \n[15,16], New Zealand [17], and southern Africa [18]. Thus, a lack of transparency about \nresearch goals, limited community engagement, misuse of samples, and failure to address \ncommunity priorities have hampered participation in past genetic studies and continues to make \nmany communities wary of participation today [19]. Within this context, several Indigenous \ncommunities and scholars have published strategies and recommendations for best practices, \nas well as explicit rules of engagement [20–23]. For example, in response to concerns about \nlack of informed consent, use of culturally-inappropriate language, and inadequate ethics review \nof research using their genetic data [18], the San people in southern Africa developed a code of \nresearch ethics. This code is centered around five broad principles — respect, honesty, justness \nand fairness, care, and process — and all research projects are reviewed against this code \nbefore they are approved [23]. At its core, this set of principles, as well as others that have been \nput forth [20,21], center around meaningful and continued engagement with participant \ncommunities during the research process. This engagement is predicated on effective \ncommunication of the science goals and processes to ensure communities understand the \nresearch being conducted, as well as its benefits and limitations.  \n Despite the acknowledged need to clearly communicate study goals, procedures, and \nresults to participant communities, it can be difficult to discuss genetics topics, which are often \ncomplex, abstract, and tied to field-specific background knowledge and jargon [24]. For \nexample, most concepts and processes in genetics are not visible to the naked eye, often \nmaking them less intuitive. Further, certain technical words that are used in English, such as \n“DNA”, “gene”, or “chromosome”, may lack direct equivalents in other languages [25]. Even \nsome terms used to describe family relationships, such as “aunt” or “cousin”, may not have \ndirect translations or may refer to different types of relationships in different cultures and \nlanguages [26]. Quantitative studies of genetics literacy further demonstrate that understanding \nof core genetics concepts varies widely across populations and can shift over time [27]. To \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\ncombat these complexities, one way scientists have communicated genetics topics to both the \ngeneral public and participant communities is through the use of images. \nImages are one of the choice methods to convey genetics material because they can \npotentially be generalizable across cultures and languages, depict processes that may not be \nvisible to the naked eye, and can be a starting point for concept sharing [28]. However, there are \nno guidelines or widely adopted models for how images should be developed or shared with \nparticipant communities. While a few examples have been published—specifically illustrations \nused for returning results [29,30] or as a supplement to informed consent [31]—these examples \ncan be quite technical and include large amounts of field-specific terminology. Those that are \nmore accessible rely on analogies, which can help provide visualization of abstract concepts by \nhighlighting similarities to familiar phenomena [32]. For example, Arango-Isaza et al. used \ncolored corn, which is an important crop with a long history of cultivation among the Mapuche \ncommunities they worked with, to explain genetic diversity and heritability [30]. Despite ongoing \nprogress in this area, a remaining gap is that there is very little information in the literature about \nhow images are developed, especially information about iterative engagement with communities \nand their requests and feedback. As a result, it also remains unclear whether engagement with \nand feedback on these materials are heterogeneous across audiences, and what \ncommunity-specific or contextual factors are important to consider. \nHere, we address these gaps by conducting a qualitative and quantitative, \ncommunity-driven study to understand how illustrations can be useful to Indigenous \ncommunities interested in learning more about genetics, with the broader goal of improving \nengagement with genetics research and promoting collaborative approaches. To do so, we \nworked with two groups with which we have long-standing relationships through ongoing \nanthropological, genomic, and biological research studies: the Orang Asli, the Indigenous \npeoples of Peninsular Malaysia, and the Turkana, Indigenous pastoralists of northwest Kenya. \nWe conducted this project in two phases. First, we created illustrations that address commonly \nasked questions about genetics from subsistence-level communities. The same set of images \nwas initially piloted with both Orang Asli and Turkana community members across multiple \nrounds of fieldwork, and feedback was iteratively collected to update the illustrations. Although \nour initial goal was to develop broadly generalizable images appropriate for both Turkana and \nOrang Asli, this process revealed the importance of population-specific imagery and framing, \nleading us to prioritize community-tailored illustrations over generalizable illustrations (and thus \nfocusing on Orang Asli; Turkana-specific images were not developed). Next, we presented the \nfinalized, population-specific illustrations to Orang Asli communities and conducted interviews \nabout the illustrations to evaluate their effectiveness. Finally, we explored interview responses to \nidentify patterns in participants’ understanding and feedback, as well as the social and \ncontextual factors shaping participants’ responses. Overall, this study responds to \ncommunity-expressed interests in learning more about genetics and offers experiences for \nresearchers seeking to engage in similar communication efforts. \n \nMethods \n \nParticipant populations  \nOrang Asli \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\n The Orang Asli are the Indigenous peoples of Peninsular Malaysia, comprising less than \n1% (~210,000 individuals) of the country’s population. They are typically divided into 19 distinct \nethnolinguistic groups and three broad sub-groups, distinguished primarily by language, \nphenotype, and subsistence strategies [33]. Data for this study were collected from ten villages \n(Figure 1A) that are primarily situated in remote regions in the rainforest of Peninsular Malaysia \nand have historically experienced limited access to infrastructure, including formal education, \ntransportation, and healthcare. The villages were predominantly occupied by members of the \nBatek, Jahai, Temiar, and Semai ethnolinguistic groups. The Batek and Jahai belong to the \nNegrito (Semang) sub-group, traditionally nomadic hunter-gatherers who speak Northern Aslian \nlanguages, while the Temiar and Semai belong to the Senoi sub-group, traditionally practicing \nswidden agriculture with a sedentary or semi-nomadic lifestyle and speaking Central Aslian \nlanguages [34]. \n Over the past 50 years, Malaysia’s rapid socioeconomic development has led to major \nlifestyle shifts for the Orang Asli, driven by two main forces: 1) the expansion of plantation \nagriculture and natural resource extraction has fragmented Orang Asli lands, and 2) government \nprograms promoting assimilation into Malaysian society have shifted many Orang Asli away \nfrom traditional villages into organized resettlement schemes. The Orang Asli Health and \nLifeways Project (OA HeLP) [35] is an international team of scientists, physicians, and Orang \nAsli advocates focused on how these lifestyle transitions are influencing health outcomes, using \na range of questionnaire, anthropometric, biomarker, and genomic data types [36–39]. \n Data were collected from consenting adults ages 18 and older during trips conducted in \npartnership with OA HeLP, including 1) mobile clinics that visit communities to conduct research \nand provide free healthcare, or 2) follow-up trips to communicate results with study \ncommunities. However, participation in either concurrent or previous OA HeLP research and \nmobile clinics was not a requirement to participate in this study. Orang Asli communities \nincluded in this project were identified through existing relationships developed over many years \nof prior work by members of the OA HeLP team. The process of recruitment included two \ngeneral steps. Permission to conduct the research was first sought from community leaders. \nThis step was followed by the individual-level consent process, where the goals, research \nquestions, and methods of this project were explained in detail after which formal, written \nconsent was given. \n  \nTurkana \nThe Turkana are a nomadic pastoralist population living in the Turkana Basin in \nnorthwest Kenya. This study worked with Turkana individuals from four villages, two of which \nwere remote and two of which were in more urbanized areas. Ongoing infrastructure \nconstruction and rapid economic development of Kenya has resulted in the growth of several \nurban centers in and near traditional Turkana lands, the expansion of small-scale markets, and \nan increased reliance on industry and agriculture. As a result, many Turkana no longer \nexclusively practice traditional pastoralism, instead relying on trade, small scale farming, and \nincreasing participation in the market economy. In addition to socioeconomic changes \nhappening within the Turkana region, many Turkana have moved to highly urbanized areas in \ncentral Kenya in the last several decades [40,41].  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\n \nFigure 1: Study overview. (A-B) Map showing the locations where illustrations were piloted in Kenya and Malaysia \nrespectively, with different colors representing the year the village was visited and shape representing the phase of \nthe study. Phase 1 indicates piloting of images occurred at the marked location and Phase 2 indicates final \npresentations of the images and structured interviews at the marked location. (C-F) Development of the illustration \nanswering the frequently asked question “What is DNA?”, where (C) is the first version presented and (F) is the final \nversion presented. Although writing here is shown in English, it was written in Malay or Swahili in the presented \nimages. When presented to participant audiences, Figure E would include photographs of Orang Asli. \n \nThe Turkana Health and Genomics Project (THGP) is an international team of \ngeneticists and biologists working to understand the health consequences of these transitions \nusing integrated questionnaire, anthropometric, biomarker, and genomic data [37,42–44]. In \npartnership with THGP, informal interviews were collected from consenting adults ages 18 and \nolder during either 1) mobile clinics that visit communities to conduct research and provide free \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\nhealthcare, or 2) community engagement and outreach events. Participation in either concurrent \nor previous THGP research and mobile clinics was not a requirement to participate in this study. \nThe study goals, research questions, and methods of this project were explained to participants \nin a language commonly used and understood within each study population by researchers \nbefore formal, written consent was given. \n \nOverview of project phases \nThe study protocol was reviewed by the BRANY Institutional Review Board (protocol no. \n24-180-734) and was determined to be exempt. The OA HeLP was approved by the Medical \nReview and Ethics Committee of the Malaysian Ministry of Health (protocol ID: \nNMRR-20-2214-55565), the Malaysian Department of Orang Asli Development (permit ID: \nJAKOA.PP.30.052 JLD 21), and the Institutional Review Board of Vanderbilt University (protocol \nID: 212175). The THGP was approved by Vanderbilt University (protocol ID: 00000162) and \nKenya Medical Research Institute (KEMRI/SERU/CTMDR/119/4875).  \nData collection occurred in two phases to develop illustrations and evaluate their \neffectiveness as a resource for communicating information about genetics research. In the first \nphase (Turkana-2022, Orang Asli-2023, Turkana-2023, Orang Asli-2024; Figure 2), we \ndeveloped initial illustrations from a list of frequently asked questions (see below: Initial \nformulation of images), after which informal interviews were conducted to identify ways to \nimprove the illustrations, using a mix of open-ended and multiple choice questions. In the \nsecond phase (Orang Asli-2025; Figure 2), we focused on updated, Orang Asli specific images \nand conducted structured interviews to assess participant response to the illustrations using \nyes/no, short-form open-ended, and ranking questions (see Supplementary Material). Finally, \nwe analyzed the interview data collected in the second phase to identify patterns in participant \nfeedback and responses. \n \nInitial formulation of images (Phase 1) \n To design the initial illustrations, we began with a list of frequently asked questions \nraised by participants in a separate long-term anthropological and health research study with \nsimilar goals to OA HeLP and THGP (namely, the Tsimane Health and Life History Project [11]).  \nThis list of questions was the focus of project team and community discussions during follow-up \nresults return trips for a recently published paper on Tsimane genetics [45]. We selected seven \nof these frequently asked questions to guide the development of the images: 1) What is DNA?, \n2) Can DNA affect your health?, 3) What can scientists learn from DNA?, 4) Besides DNA, what \nelse can you find from my blood?, 5) Why are scientists interested in markers of health in \nblood?, 6) What happens to my blood once you collect it?, and 7) Who has access to my DNA?  \n We created nine generalizable images to illustrate the above questions. To \naccommodate variable literacy levels, we minimized written text. All writing included on \nillustrations piloted in Turkana was written in Swahili, and writing on illustrations piloted with \nOrang Asli was written in Malay. Although Swahili and Malay are not the traditional languages \nwithin these groups, they function as regional lingua francas and are widely spoken in each \narea. We solicited feedback from community members and field assistants during two field \nseasons in Kenya (Turkana-2022, Turkana-2023) and two field seasons in Malaysia (Orang \nAsli-2023 and Orang Asli-2024). We used an iterative approach to incorporate community  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\n \nFigure 2: Timeline of illustration development, piloting, and evaluation across study phases. Key milestones in \nthe iterative creation of genetics illustrations, including initial identification of frequently asked questions, piloting of \ngeneralizable illustrations, refinement of presentation formats, and the transition to Orang Asli–specific illustrations \nprior to formal interview-based evaluation. \n \nfeedback, updating the images during each field season (Figure 1C). Ultimately, two versions of \nthe illustrations were produced -- one generalizable version suited for multiple contexts (see \nhttps://github.com/audreyarner/genetic_illustrations), and one Orang Asli-specific version \n(https://github.com/tcmccabe/OrangAsliHealthIllustrations).  \n \nPresentation of illustrations (Phase 1 and Phase 2) \n In Phase 1, we piloted the dissemination of the illustrations using multiple \nformats--including live slide presentations, pre-recorded videos, tablet-based viewings, and \nprinted discussion-based materials--to refine what was most useful to participants and could be \nreliably implemented, given that some locations do not have consistent access to electricity, wifi, \nor cell service. Similar to the illustrations themselves, the presentation of the illustrations was \nupdated in an iterative manner, incorporating feedback to identify the most useful approach.  \nThe illustrations were first disseminated at a Turkana community engagement event held \nin October 2022 (Turkana-2022). They were presented as a live talk, where a THGP research \nassistant presented the material in Swahili. Each illustration was pictured on a PowerPoint slide, \nand was shown via a projector. Later in the same field season (Turkana-2022), we presented a \nrevised set of images into a 15-minute slide presentation that was narrated in Swahili and \nshown to individuals in three villages during mobile health clinics. The video format ensured a \nconsistent presentation each time. The video format was also presented at the 2023 Turkana \nCultural Festival, where THGP team members hosted a booth that highlighted their broader \nresearch activities and presented the illustrations (Turkana-2023). \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\n We also presented the images in the 15-minute video format to Orang Asli communities \n(Orang Asli-2023). In this case, the video was pre-recorded with explanations in Malay and \nshown to community members in either large (~20 people) community gatherings using a \nprojector or small (2-5 people) group settings on a tablet (Figure 1B). \nBased on feedback, we updated the presentation format such that OA HeLP research \nassistants led small discussion-based presentations of printed images, explaining each image to \ngroups of one to six individuals (Orang Asli-2024). Each illustration was printed and laminated \non A4 paper. Although the information conveyed varied slightly between sessions, this format \nallowed for interactive discussion and hands-on engagement with the materials. We engaged a \nbroad cross-section of community members in these discussions, including Tok Batin (village \nheadmen), teachers, community elders, and young adults who had recently completed \nsecondary schooling. To ensure accessibility beyond digital settings and as a future resource, \nwe also produced printed pamphlets featuring the same content (Orang Asli-2024). \nIn Phase 2 (Orang Asli-2025), the final illustrations were shown during a live \npresentation in a medium-large group setting to evaluate efficacy and gather participant \nfeedback. Similar to earlier formats, this presentation used either PowerPoint slides (when \nelectricity was available) or laminated, printed versions for the participants to view. These \npresentations were delivered live in Malay by an OA HeLP research assistant, which allowed for \ninteractivity and interruptions if viewers had questions. The research assistant involved in Phase \n2 did not have a background in genetics, but instead had discussed the image explanations with \na researcher who had a background in the field to improve phrasing and understandability. The \nfinal presentations lasted approximately 12 minutes. The presentation was delivered in six \nOrang Asli communities, with between 20 and 70 community members attending each session. \nWe again provided pamphlets featuring the same content to participants for future reference.  \n \nInterviews (Phase 1 and Phase 2) \n In Phase 1, we conducted brief, unstructured interviews during four field seasons \n(Turkana-2022, Orang Asli-2023, Turkana-2023, Orang Asli-2024) to assess how the \nillustrations, their presentation, and the interview itself could be improved to best meet \ncommunity needs. Interviews were conducted in either Malay with Orang Asli participants or \nSwahili with Turkana participants, and individuals were free to answer whichever questions they \nwanted. Questions included open-ended items about what individuals liked most and least \nabout the images, as well as gain of knowledge questions assessing understanding of some of \nthe illustrated genetics concepts. \nIn Phase 2, we conducted structured, short format interviews with Orang Asli participants \nto collect both qualitative and quantitative data (Orang Asli-2025). These interviews focused on \nfour main areas: demographic information, prior knowledge before viewing the presentation, \nopinions of the illustrations, and perceived knowledge empowerment (see Supplementary \nInformation). Several questions were refined from those piloted during earlier field seasons. All \ninterviews were conducted in Malay by local OA HeLP research assistants and lasted \napproximately 10-15 minutes. In total, 92 participants across six villages completed the \nstructured interviews (SI Table 2). \n \nThematic analysis of responses to open-ended questions (Phase 2) \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\nTo identify broad ideas underlying responses to the three open-ended interview \nquestions, we conducted an iterative, inductive thematic analysis [46] of participant responses \nto each question separately. Specifically, two researchers (A.M.A. & A.S.) independently \nopen-coded all responses to each question inductively using MAXQDA version 26 software to \nidentify recurring concepts and patterns. A given response could have multiple phrases \nconveying different ideas. Therefore, the unit of analysis in coding was a phrase connected to \nan idea. Initial coding of the responses to each interview question was discussed, where \nresearchers systematically reviewed code definitions and applications, with discrepancies \nresolved by consensus. Percent agreement was calculated to assess coding consistency \nbetween researchers, which ranged from 86% to 100% agreement (SI Table 3). After consensus \ncoding was reached, researchers independently identified higher-order themes that emerged \nfrom the identified codes, followed by reflexive discussion to clarify and refine thematic structure \nand description for each theme. The most common themes were identified based on number of \nmentions (SI Figures 1-3). \n \nStatistical analysis of interview data (Phase 2) \n First, we used binomial models to test whether the proportion of individuals responding \naffirmatively to each yes/no question differed from that expected by chance, fitting models \nseparately for each question. We corrected for multiple hypothesis testing using a \nBenjamini-Hochberg false discovery rate [47]. We also calculated a pairwise Pearson correlation \nmatrix to evaluate how responses to each question co-varied across individuals. \nSecond, we tested whether answers to the questions in our interviews could be \ncomposed into delineable axes of variation (e.g., whether groups of individuals tended to \nanswer certain questions similarly). To do so, we used a multiple correspondence analysis \n(MCA), a type of exploratory factor analysis designed to reduce the dimensionality of categorical \ndata [48]. Our MCA included the pool of the eight yes/no questions converted to Boolean format \n(true/false). Two individuals were removed due to one or more missing answers, resulting in a \ntotal of 90 individuals for this analysis. MCA was performed using the FactoMineR package in R \nwith default parameters [49]. The first two dimensions were retained based on their relative \ninertia (35.3% and 18.9% respectively; see SI Figure 4) and interpretability.  \nWe next tested whether individual coordinates on MCA dimensions 1 and 2 were shaped \nby any sociodemographic factors, namely sex, age, highest education level (coded as a linear \nvariable with 0 representing no formal education, 1 representing some primary education, 2 \nrepresenting some secondary education, and 3 representing some university education), and \n“urbanicity”. Here, we used a location-based “urbanicity score” that captures access to \nindustrialized, market-based resources available across the community (e.g., access to \nelectricity, sewage, formal education; see Supplementary Text for urbanicity score generation). \nThis score was first proposed by Novak et al [50] and has previously been tested in Orang Asli \n[36]. We used linear models including sex, age, highest education level, and urbanicity score to \npredict MCA dimensions 1 and 2 in separate models [47]. We also ran follow-up models in \nwhich highest education level was coded as a binary variable of no formal education (coded as \n0) versus any level of formal education (coded as 1). \nFinally, we used linear models to analyze whether demographic and other factors \nimpacted response to each yes/no question. For each question separately, we fit a binomial \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\nmodel in which response (yes/no) was predicted jointly by age, sex, highest level of schooling, \nor urbanicity score. We again corrected for multiple hypothesis testing using an FDR approach. \nSimilar to the above, we ran follow-up models switching highest education level with a binary \nvariable of no formal education versus any level of formal education. All analyses were \nperformed using the R computing language and RStudio (version 4.2.1).   \n \nResults \n \n We developed a series of illustrations to address frequently asked questions about \ngenetics. Both a broad, generalizable version and an Orang Asli-specific version of the \nillustrations are available on GitHub (https://github.com/audreyarner/genetic_illustrations, \nhttps://github.com/tcmccabe/OrangAsliHealthIllustrations) and are also accessible from the OA \nHeLP project website (https://www.orangaslihealth.org/). In the following text, we describe: \nPhase 1, which included the iterative development and refinement of the illustrations; and \nPhase 2, which included the presentation of the final version of illustrations and the qualitative \nand quantitative evaluations of their efficacy and drivers of engagement.  \n \nPhase 1: Iterative development of illustrations depends on community feedback \n Given the desirability of a generally-applicable genetics resource, our first round of pilot \nimages depicted people, objects, and environments that were not specific to any geographic \nregion (Figure 1C). For example, we used a simplified human outline without any identifiable \nphenotypes or characteristics to enhance relatability across contexts, consistent with genetics \nimagery used in prior publications [29,30]. A key theme in early feedback (Turkana-2022) was \nthe desire for more realistic images. In response, we revised the illustrations to include greater \nvisual detail and less abstract depictions of individuals, including a range of skin tones (Figure \n1D). Feedback on the revised illustrations was generally positive; however, viewers found the \nmode of illustration dissemination (a video; Turkana-2023, Orang Asli 2023), to be too long and \ninsufficiently interactive. Therefore, participants suggested the inclusion of more dynamic \nelements such as animation. Similar to feedback from Turkana, Orang Asli viewers suggested \nthat the video was too long, with viewers reporting that this style of presentation was not \ninteractive enough (Orang Asli-2023). Orang Asli feedback also provided new perspectives, \nhighlighting a desire for the presentation to include imagery that was more locally relatable and \nspecific to their lives.  \nBased on this community feedback, we prioritized shortening the presentation, making \nthe presentation more interactive, and incorporating Orang Asli-specific elements into the \nillustrations for additional piloting (Figure 1E, Orang Asli-2024). To shorten the presentation we \nremoved one of the images (which answered the question “What happens to my blood once you \ncollect it”) that was the most repetitive. For each image, we included relevant pictures taken in \nOrang Asli communities, as well as added a rainforest background to each of the illustrations. \nThe most common suggestion we received at this stage was to incorporate additional Orang \nAsli-specific examples and imagery, as well as examples of genetics principles that community \nmembers would be more familiar with. Although our original goal was to produce a \ngeneralizable resource suitable for multiple populations, this feedback prompted a pivot toward \ndeveloping an Orang Asli-specific version as the primary set of illustrations.  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\n Feedback also informed the development of our interview questions. During early pilot \ninterviews (Turkana-2022, Orang Asli-2023), some community members noted that it felt like \ntaking a test, which they remarked was not enjoyable. Because our goal was to assess how \neffective the illustrations were for knowledge empowerment rather than acquisition by Western \nstandards, we removed items with definitive “right” or “wrong” answers. Additionally, we limited \nthe number of open-ended questions included in the final interview, as we found that individuals \nhad a difficult time putting some concepts into words without prompts. Overall, community \nfeedback was consistently constructive, emphasizing appreciation for the visual and oral \napproach and the perceived increasing relevance of the illustrations to their own experiences. \n \nPhase 2: Final content and presentation focused on population-specific imagery \n The finalized illustrations were structured around six frequently asked questions about \ngenetics (Figure 3; SI Table 1; Orang Asli-2025). Guided by community recommendations, this \nOrang-Asli specific version incorporated recognizable local examples of genetics. For instance, \nhair texture -- which is highly variable within Orang Asli ethnolinguistic groups -- was used to \nillustrate heredity (Figure 3B), replacing height which shows less visible local variation. \nAdditionally, we used durian, a popular fruit in Malaysia that has many easily-recognizable \nvarieties varying in appearance, texture, and taste, to explain genetic diversity and the impact \nthe environment can have on traits (Figure 3C). Because small-scale farming and close \ninteraction with cultivated and wild plants are a part of daily life for many Orang Asli \ncommunities, this example leveraged shared experiential knowledge to make genetic variation \nmore accessible. Similar locally grounded adjustments were made across all illustrations, \nincluding depicting people in traditional clothing and housing.  \n \n \nPhase 2: Illustrations reported as useful by participants  \n To understand whether the illustrations were helpful in relaying genetics concepts, we \nconducted structured interviews with 92 Orang Asli individuals (Supplementary Table 2). In total, \n85 participants (92%) reported wanting to know more about genetics research, and 44 \nparticipants (48%) reported they had believed prior to seeing the illustrations that there was \nhealth-related information in their blood. Participants answering affirmatively to the second \nquestion were asked a follow-up about the type of information they believed would be present. \nNine individuals did not have a specific response. For those who responded, we used a \nthematic analysis to identify two major themes that developed from the data (SI Figure 1). First, \nparticipants described knowledge of measurable indicators coming from blood, often referencing \nspecific biological markers or tests (e.g., “blood has sugar in it”). Second, participants expressed \nknowledge that blood can be used to assess medical status, reflecting broader health \ninterpretations (e.g., “diseases and health”). \n \nQualitative analysis: Illustration preferences align with familiarity, while technical images are \nmore confusing   \nWe then asked questions about the illustrations to understand participants’ preferences \nand points of confusion. The greatest percentage of participants (38%) reported Illustration B as  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\n \nFigure 3: Final, Orang Asli-specific illustrations presented to communities. Below each image, we specify the \nfrequently asked question being depicted. Throughout the text and in other figures, we refer to each image by their \nassociated letter here (A-H). When presented, Illustration E included photographs of other Indigenous populations \nfrom the locations specified. \n \ntheir favorite (Figure 3B). This image is one that Orang Asli would likely be the most familiar \nwith, depicting the inheritance of hair texture -- a trait with observable variation among Orang \nAsli -- showing individuals residing with their family in traditional bamboo houses set in a \nrainforest environment. To formally assess why certain images were preferred, we conducted a \nthematic analysis of participants’ open-ended explanations for their favorite images (SI Figure \n2). The most common theme (n=54 mentions) was preference for imagery related to identity, \nwith participants frequently referencing recognizable environments and lived experiences, \nnoting, for example, that the illustration “is the same as my daily activities, like playing takraw”. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\nParticipants also emphasized an interest in genetics concepts (n=31 mentions), explaining that \nthey liked some images because they “liked knowing that everyone has their own DNA”. Smaller \nsubsets of participants expressed preference for the visual and aesthetic components (e.g., “the \npicture is beautiful and elegant”; n=10 mentions), health-related imagery (e.g., “because the \npicture shows how you can be healthy”; n=25 mentions), and depictions that there are \nlifestyle-related health benefits (e.g., “healthy lifestyle of the village”; n=25 mentions). \nWe also asked which images, if any, were confusing to viewers (Figure 4B); 85% of \nparticipants reported at least one image as confusing (mean=1.8 confusing images). For \nexample, the image most often reported as confusing (selected in 30% of all image responses, \nwith participants able to select multiple images) was Illustration E, which depicts DNA variation \nin other Indigenous communities. This image was the most technical; however, it was retained \nin the final set of illustrations given feedback from Phase 1 to include information about other \nIndigenous populations around the world.  \nIn order to understand topics of continued interest, we asked what images, if any, \nparticipants would want to learn more about. Most participants reported wanting to learn more \nabout at least one image (mean=1.4 images). Interest was fairly evenly distributed across the \ndifferent illustrations (Figure 4C). As a follow up, we asked what general topics participants \nwould like to have learned more about (SI Table 4). Most participants expressed interest in \nlearning more about health and disease (49% of individuals) and relatedness (46% of \nindividuals). A small number of participants selected “other”, primarily raising questions about \nblood type.  \nFinally, we asked participants what one thing they learned from the illustrations was. We \nagain used a thematic analysis of short-form open-ended responses, which revealed four main \nthemes (SI Figure 3). The most common theme reflected increased understanding of the role of \nDNA and blood in the body, with participants describing new awareness that blood contains \nbiological information and that DNA influences bodily traits and health (n=73 mentions). For \nexample, some participants noted “everyone has their own DNA” and “DNA changes can impact \nhealth”. A second theme involved factors contributing to health and well-being (n=32 mentions). \nResponses referenced learning, for example, that “blood has health information.” The third \ntheme captured recognition of genetic variation among individuals and populations (n=21 \nmentions). Finally, a smaller but important theme reflected awareness of a knowledge gap, with \nparticipants identifying difficulty articulating a specific concept they learned or noting they \nwanted to learn more in the future (n=7 mentions).  \n \nQuantitative analysis: Genetics illustrations were broadly engaging and improved \nunderstanding, and engagement showed modest variation by education, sex, and urbanicity \nWe sought to understand the effectiveness of the illustrations for participants' wants by \nasking eight yes/no questions assessing self-reported interest and knowledge gain. All \nquestions were answered affirmatively more than expected by chance (FDR<0.05), indicating \nthat participants found the illustrations engaging, understandable, and informative more so than \nnot (Figure 6A, SI Table 5). We next assessed the correlation between answers (Figure 6B). \nResponses showed internal consistency, with measures of self-reported understanding and \nengagement positively correlated with one another (e.g., “I would view the images again” and “I \nwould recommend to a friend”). Reporting that at least one of the illustrations was hard to  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\n \nFigure 4. Participants’ preference and interests in genetics illustrations. Illustrations labeled according to Figure \n3. (A) Barplot of participants’ favorite image. Each participant could select only one illustration as their favorite. (B) \nBarplot indicating which images were confusing. Participants could select multiple images (C) Barplot showing the \nnumber of images each participant reported as confusing. (D) Barplot indicating which images participants wanted to \nknow more about. (E) Plot depicting the number of images each participant chose, with multiple illustrations able to \nbe chosen. \n \nunderstand was negatively correlated with nearly all other questions, particularly questions \nrelated to engagement and recommendation to others (mean Pearson r = -0.2).  \nTo further explore correlations in participants’ responses, we performed a multiple \ncorrespondence analysis (MCA) using all eight yes/no questions. We determined that two \ndimensions accounted for the majority of the variance (dimension 1 proportion of variance: \n35.3%, dimension 2 proportion of variance: 18.9%; SI Figure 1). The first dimension appeared to \nprimarily reflect interest and engagement, with high loadings for items such as “I would \nrecommend the illustrations to a friend” and “I want to learn more” (Figure 6C). The second \ndimension was driven by questions related to understanding and clarity, loading more strongly \non questions such as “the illustrations helped me understand more about genetics” and “I \nunderstand why scientists would want to study DNA” (Figure 6D). We then modeled \nassociations between individuals’ scores on the first two MCA dimensions and four predictors: \nage, sex, highest attained level of education, and urbanicity. Although none of the predictors \nremained significant after multiple hypothesis testing correction, dimension 2 was nominally \nassociated with urbanicity score (p=0.042) (SI Table 6).  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\nFinally, we modeled each individual interview question as a function of age, sex, highest \nattained level of education, and urbanicity (Figures 6E, SI Table 7). While no predictors \nremained significant after multiple hypothesis testing correction, we found that highest education \nlevel showed the most consistent associations, reaching a nominal P<0.05 for three questions. \nA higher level of education was associated with “yes” responses for all three of these questions \n(Figure 6F). Additionally, we found that individuals with lower urbanicity were more likely to \nreport that the illustrations helped them understand why researchers want to study DNA \n(P=0.03; SI Table 7), while individuals with lower urbanicity were more likely to report that they \ncould explain the illustrations to a friend (P=0.003; SI Table 7). Finally, men were more likely \nthan women to report that some of the illustrations were hard to understand (P=0.01; SI Table \n7). We found that results were very similar when using a binary of none vs any formal education \n(SI Table 8). Given the importance of specific histories, we ran additional models including \nethnolinguistic group as a predictor. Together, these patterns suggest that prior exposure to \nformal education and biology concepts, which vary systematically with urbanicity, can influence \nhow individuals interpret and engage with genetics communication materials.  \n  \nDiscussion \nEffective communication of genetics information is essential to ethical research \npartnerships [20,23]. However, genetics concepts are often abstract, technical, and challenging \nto convey across diverse linguistic and cultural contexts. Visualization strategies can bridge this \ngap by connecting intangible concepts to concrete, visible examples. While a few examples of \nillustrations [29–31] and videos [51,52] explaining genetics concepts to participant communities \nhave been published, there is limited information about how these visuals are developed or \nreceived. Here, we used an iterative, community-based process to demonstrate that illustrations \ncan serve as effective tools for communicating genetics research with Indigenous communities, \nbut that engagement and comprehension are shaped by demographic and contextual factors. \nFirst, we found that iterative, community-driven development was essential for producing \nillustrations that aligned with participant priorities. The initial images we created (Figure 1C) \ndiffered substantially from the final versions (Figure 1F), with multiple rounds of revision \ninformed by community feedback. This process aligns with participatory research approaches \nthat emphasize co-creation and responsiveness to users rather than one-directional knowledge \ntransmission. Indeed, prior work in science communication and community-based participatory \nresearch has shown that iterative development improves relevance, trust, and engagement, \nparticularly when communicating complex or sensitive topics [53,54]. Interestingly, this iterative \nprocess revealed that participant priorities did not always align with maximizing simplicity or \nimmediate clarity. For example, although most participants reported at least one illustration as \nconfusing (Figure 4C), the image most frequently identified (Figure 3E, depiction of genetic \nvariation across Indigenous populations globally) was intentionally retained following participant \nfeedback emphasizing the importance of understanding how local communities fit within a \nbroader Indigenous context, although our study cannot fully disentangle whether this confusion \narose from the visual representation itself or from the underlying concept being communicated. \nSecond, we found that participants strongly preferred illustrations that reflected familiar \npeople, environments, and lived experiences. During early phases of development, both \nTurkana and Orang Asli participants expressed dissatisfaction with illustrations that used  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\n \nFigure 6. Self-reported satisfaction is high for community-specific genetics illustrations. (A) Questions \nparticipants were asked. Their number of 1 through 8 is repeated across panels of the figure. (B) Barplot showing the \npercentage of participants who answered “yes” vs “no” for each question used to determine effectiveness of \nillustrations. Binomial modeling was used to test whether each proportion was significantly different than 0. (C) \nCorrelation between participant answers to each question. Numbers correspond to the questions in panel A. (D-E) \nLoadings of each question in MCA for dimensions 1 and 2 respectively. (F) Forest plot showing effect size and \nconfidence interval for each demographic covariate of interest using binomial modeling. Color represents significance \nthreshold.  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\ngeneric human figures or attempted to represent diversity through a range of skin tones; \ninstead, participants wanted to see individuals who looked like themselves and contexts that \nreflected their own communities. This finding aligns with prior work demonstrating that analogies \ngrounded in shared experiences facilitate comprehension of abstract biological concepts. For \nexample, digital storytelling in Alaska Native communities has been shown to be a culturally \nrespectful and engaging approach to science communication, particularly when narratives are \nrooted in local knowledge and experience [55]. More broadly, this finding mirrors patterns \nobserved in ethical governance frameworks: while global principles such as the United Nations \nDeclaration on the Rights of Indigenous Peoples (UNDRIP) and the CARE principles [56,57] \nprovide important guidance, their effective application requires attention to the specific histories, \ncultures, and priorities of individual Indigenous communities, which are not homogeneous \n[23,58,59]. \nFinally, the heterogeneity we observed in engagement with and understanding of the \nillustrations further reinforces the need to move beyond a generalizable, “one size fits all” \napproach. Although participants overall showed significant interest in learning about genetics, \nstrong engagement, and perceived gains in understanding, our quantitative analyses revealed \nmodest but consistent association between responses and urbanicity, education, age, and sex. \nEducation level in particular emerged as a key predictor, likely reflecting differential exposure to \ngenetics-related topics. Individuals with more formal education were more likely to report that \nthey could explain the illustrations to a friend, potentially reflecting greater prior familiarity, but \nalso that they would like to learn more about these topics, suggesting that prior educational \nexposure may also foster greater confidence and interest in scientific information. Additionally, \nindividuals in less urbanized locations were more likely to report that the illustrations helped \nthem understand why researchers wanted to study DNA, consistent with this being an early \nexposure to these concepts. Furthermore, we observed sex-based differences in engagement \nand interpretation, suggesting that learning preferences and perceived relevance may vary \nacross social roles and experiences, a phenomenon identified in previous literature [60,61]. \nOur study has several limitations. First, while illustrations were piloted with both Turkana \nand Orang Asli communities, our decision to switch to community-specific illustrations resulted \nin formal evaluations only being conducted with Orang Asli participants, making us unable to  \ncompare illustration effectiveness across populations. Second, interviews relied on retrospective \nself-assessment rather than objective baseline measures, which may introduce recall bias [62]. \nFinally, because the final version of illustrations were delivered as live presentations by OA \nHeLP research assistants, slight variation in phrasing across presentations may have influenced \nresponses. However, we view this flexibility as a strength rather than a weakness, reflecting \nreal-world conditions in which engagement is relational, interactive, and adaptive, rather than \nstandardized. Additionally, because illustrations were presented alongside verbal explanations, \nwe are unable to determine the extent to which visual materials themselves contributed to \ncomprehension relative to verbal explanations alone. More broadly, visual limitations themselves \nhave inherent limitations. For example, several community members expressed interest in \nanimated versions of the illustrations, suggesting that motion and narration may further enhance \nclarity for complex biological processes. While animation would allow for more dynamic \nexplanations, producing high-quality animated materials requires substantial financial resources, \ntechnical expertise, and reliable technological infrastructure. Moreover, engagement formats \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\nmust remain concise; in our experience, community members are unlikely to engage with \nmaterials longer than 10-15 minutes. These considerations highlight the balance researchers \nmust strike between ideal communication tools and practical constraints of time, funding, and \nlocal context. Rather than serving as comprehensive explanations of research in themselves, \nillustrations are best understood as accessible entry points that can prompt further discussion \nand ongoing dialogue.  \n Despite these limitations, we hope that reporting on the challenges and procedures \ninvolved in this work offers guidance for others. We have identified three practical implications, \nwhich largely echo prior themes [19,20,30]. First, effective communication materials should be \ntreated as evolving resources rather than finalized products, with time and resources allocated \nfor iterative revision. We acknowledge (and experienced) that this can be challenging given the \ndifficulty of securing dedicated resources (e.g., grant funding) for such work. Second, \ncommunity-specific tailoring should be considered a critical aspect to illustration design. Third, it \nis important to evaluate engagement materials not only for comprehension, but also for whether \nthey resonate with participants’ interests, values, and goals for engaging with researchers. We \nhave already begun applying these principles to other OA HeLP and THGP initiatives, including \nrecent results-return efforts. An additional area for future work is to examine whether \nparticipatory communication exercises such as this influence broader trust in science and \nresearchers, particularly among individuals who may have had negative prior experiences with \nhealth projects or biological sample collection. As genomic research continues to occur \nalongside historically underrepresented, Indigenous populations, approaches like these aim to \noffer a pathway for building trust and fostering mutual understanding.  \n  \nData availability statement \n OA HeLP’s highest priority is the minimization of risk to study participants. OA HeLP \nadheres to the ‘CARE Principles for Indigenous Data Governance’ (Collective Benefit, Authority \nto Control, Responsibility, and Ethics). OA HeLP is also committed to the ‘FAIR Guiding \nPrinciples for scientific data management and stewardship’ (Findable, Accessible, Interoperable, \nReusable). To adhere to these principles while minimizing risks, individual-level data are stored \nin the OA HeLP protected data repository, and are available through restricted access. \nRequests for de-identified, individual-level data should take the form of an application that \ndetails the exact uses of the data and the research questions to be addressed, procedures that \nwill be employed for data security and individual privacy, potential benefits to the study \ncommunities and procedures for assessing and minimizing stigmatizing interpretations of the \nresearch results. Requests for de-identified, individual-level data will require institutional IRB \napproval (even if exempt). OA HeLP is committed to open science and the project leadership is \navailable to assist interested investigators in preparing data access requests (see \norangaslihealth.org for further details and contact information).  \nSummaries of all presented data are in the Supplementary Materials. Scripts used for \nthese analyses can be found on GitHub (https://github.com/audreyarner/genetic_illustrations).  \n \nAcknowledgements \n Above all else, we thank the Orang Asli and Turkana participants who have generously \nallowed us to work in their communities, as well as for their hospitality and support of this \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\nproject. We also thank members of the Orang Asli Health and Lifeways Project and Turkana \nHealth and Genomics Project who reviewed the early versions of the illustrations. We are also \ngrateful to Jada Benn Torres and the members of the Lea Lab for their feedback and support. \n \nFunding \nAMA was supported by the National Science Foundation’s Graduate Research \nFellowship Program (1937963 & 2444112) and Doctoral Dissertation Improvement Grant \n(2419584), as well as a Wenner-Gren Dissertation Fieldwork Grant, Leakey Foundation \nResearch Grant, and a Vanderbilt Award for Doctoral Discovery. We also thank the Vanderbilt \nEvolutionary Studies Initiative for their financial support. AJL, IW, and TSK were supported by \nthe National Science Foundation (Biological Anthropology 2142090). \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\nWorks Cited \n1.  Abdill RJ, Adamowicz EM, Blekhman R. Public human microbiome data are dominated by \nhighly developed countries. PLoS Biol. 2022;20: e3001536. \n2.  Breeze CE, Beck S, Berndt SI, Franceschini N. The missing diversity in human epigenomic \nstudies. Nat Genet. 2022;54: 737–739. \n3.  Popejoy AB, Fullerton SM. Genomics is failing on diversity. Nature. 2016;538: 161–164. \n4.  Romero IG, Rodenberg G, Arner AM, Li L, Beasley IJ, Rossow R, et al. Public RNA-seq \ndata are not representative of global human diversity. bioRxiv. 2024. \ndoi:10.1101/2024.10.11.617967 \n5.  Oh SS, Galanter J, Thakur N, Pino-Yanes M, Barcelo NE, White MJ, et al. Diversity in \nClinical and Biomedical Research: A promise yet to be fulfilled. bioRxiv. bioRxiv; 2015. \ndoi:10.1101/034538 \n6.  Peterson RE, Kuchenbaecker K, Walters RK, Chen C-Y, Popejoy AB, Periyasamy S, et al. \nGenome-wide association studies in ancestrally diverse populations: Opportunities, \nmethods, pitfalls, and recommendations. Cell. 2019;179: 589–603. \n7.  Fatumo S, Chikowore T, Choudhury A, Ayub M, Martin AR, Kuchenbaecker K. A roadmap \nto increase diversity in genomic studies. Nat Med. 2022;28: 243–250. \n8.  Enabling the genomic revolution in Africa: H3Africa is developing capacity for health-related \ngenomics research in Africa. The H3Africa Consortium. \n9.  Gurdasani D, Carstensen T, Fatumo S, Chen G, Franklin CS, Prado-Martinez J, et al. \nUganda genome resource enables insights into population history and genomic discovery in \nAfrica. Cell. 2019;179: 984–1002.e36. \n10.  Nagai A, Hirata M, Kamatani Y, Muto K, Matsuda K, Kiyohara Y, et al. Overview of the \nBioBank Japan Project: Study design and profile. J Epidemiol. 2017;27: S2–S8. \n11.  Gurven M, Stieglitz J, Trumble B, Blackwell AD, Beheim B, Davis H, et al. The Tsimane \nHealth and Life History Project: Integrating anthropology and biomedicine. Evol Anthropol. \n2017;26: 54–73. \n12.  Liebert MA, Snodgrass JJ, Madimenos FC, Cepon TJ, Blackwell AD, Sugiyama LS. \nImplications of market integration for cardiovascular and metabolic health among an \nindigenous Amazonian Ecuadorian population. Ann Hum Biol. 2013;40: 228–242. \n13.  Golden CD, Borgerson C, Rice BL, Allen LH, Anjaranirina EJG, Barrett CB, et al. Cohort \ndescription of the Madagascar Health and Environmental Research-Antongil \n(MAHERY-Antongil) study in Madagascar. Front Nutr. 2019;6: 109. \n14.  Mills MC, Rahal C. A scientometric review of genome-wide association studies. Commun \nBiol. 2019;2: 9. \n15.  After Havasupai litigation, Native Americans wary of genetic research. Am J Med Genet A. \n2010;152A: fmix. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\n16.  The Havasupai Indian tribe case-lessons for research involving stored biologic samples. \n17.  Merriman T, Cameron V. Risk-taking: behind the warrior gene story. N Z Med J. 2007;120: \nU2440. \n18.  Chennells R, Steenkamp A. International genomics research involving the San people. \nSpringerBriefs in Research and Innovation Governance. Cham: Springer International \nPublishing; 2018. pp. 15–22. \n19.  Garrison NA, Hudson M, Ballantyne LL, Garba I, Martinez A, Taualii M, et al. Genomic \nresearch through an Indigenous lens: Understanding the expectations. Annu Rev Genomics \nHum Genet. 2019;20: 495–517. \n20.  Claw KG, Anderson MZ, Begay RL, Tsosie KS, Fox K, Garrison NA, et al. A framework for \nenhancing ethical genomic research with Indigenous communities. Nat Commun. 2018;9: \n2957. \n21.  Genetic research in Aboriginal and Torres Strait Islander communities: continuing the \nconversation. \n22.  Tauali i M, Davis EL, Braun KL, Tsark JU, Brown N, Hudson M, et al. Native Hawaiian views \non biobanking. J Cancer Educ. 2014;29: 570–576. \n23.  Callaway E. South Africa’s San people issue ethics code to scientists. Nature. 2017;543: \n475–476. \n24.  Yu H. Encoding complex data in popular science genetics illustrations. Inf Des J. 2014;21: \n189–206. \n25.  Kumanda N, Chiphambo S, Mtsi N. Teaching of Natural Sciences concepts to English \nsecond language speakers in primary schools in South Africa. Education and Human \nDevelopment. IntechOpen; 2022. \n26.  Shaw A, Ahmed M. Translating genetics leaflets into languages other than English: lessons \nfrom an assessment of Urdu materials. J Genet Couns. 2004;13: 321–342. \n27.  Little ID, Koehly LM, Gunter C. Understanding changes in genetic literacy over time and in \ngenetic research participants. Am J Hum Genet. 2022;109: 2141–2151. \n28.  Wood KD, Tinajero J. Using pictures to teach content to second language learners. Middle \nSch J. 2002;33: 47–51. \n29.  Bankoff RJ, Perry GH. Hunter–gatherer genomics: evolutionary insights and ethical \nconsiderations. Curr Opin Genet Dev. 2016;41: 1–7. \n30.  Arango-Isaza E, Aninao MJ, Campbell R, Martínez FI, Shimizu KK, Barbieri C. Bridging the \ngap: returning genetic results to indigenous communities in Latin America. Front Genet. \n2023;14: 1304974. \n31.  Rodriguez JJRB, Cuales JMD, Herrera MJB, Zubiri LAM, Muallil RN, Ishmael AI, et al. \nEthical challenges in genetic research among Philippine Indigenous Peoples: Insights from \nfieldwork in Zamboanga and the Sulu Archipelago. Front Genet. 2022;13: 901515. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\n32.  Duit R. On the role of analogies and metaphors in learning science. Sci Educ. 1991;75: \n649–672. \n33.  The Orang Asli and the Contest for Resources: Indigenous Politics, Development, and \nIdentity in Peninsular Malaysia. \n34.  Endicott K, editor. Malaysia’s original people. Singapore, Singapore: NUS Press; 2016. \n35.  Wallace IJ, Lea AJ, Lim YAL, Chow SKW, Sayed IBM, Ngui R, et al. Orang Asli Health and \nLifeways Project (OA HeLP): a cross-sectional cohort study protocol. BMJ Open. 2022;12: \ne058660. \n36.  Watowich MM, Arner AM, Wang S, John E, Kahumbu JC, Kinyua P, et al. The built \nenvironment is more predictive of cardiometabolic health than other aspects of lifestyle in \ntwo rapidly transitioning Indigenous populations. medRxiv. 2024. \ndoi:10.1101/2024.08.26.24312234 \n37.  Watowich MM, Brassington L, Longtin A, Wang S, Rossow R, Reinhardt KD, et al. \nUrbanization exacerbates age-associated declines in cardiometabolic health in Turkana \nand Orang Asli. medRxiv. 2025. doi:10.1101/2025.06.06.25329160 \n38.  Kraft TS, Wallace IJ, Lim YAL, Tam KL, Huat TBTATB, Tennyson R, et al. Physical activity \nand cardiometabolic health across an extreme lifestyle gradient. Public and Global Health. \nmedRxiv; 2025. Available: \nhttps://www.medrxiv.org/content/10.1101/2025.07.11.25331394v1 \n39.  Brassington L, Arner AM, Watowich MM, Damstedt J, Ng KS, Lim YAL, et al. Integrating the \nThrifty Genotype and Evolutionary Mismatch Hypotheses to understand variation in \ncardiometabolic disease risk. Evol Med Public Health. 2024;12: 214–226. \n40.  Akall G. Effects of development interventions on pastoral livelihoods in Turkana County, \nKenya. Pastoralism. 2021;11: 23. \n41.  Lea AJ, Martins D, Kamau J, Gurven M, Ayroles JF. Urbanization and market integration \nhave strong, nonlinear effects on cardiometabolic health in the Turkana. Sci Adv. 2020;6. \ndoi:10.1126/sciadv.abb1430 \n42.  Lea AJ, Martins D, Kamau J, Gurven M, Ayroles JF. Urbanization and market integration \nhave strong, nonlinear effects on cardiometabolic health in the Turkana. Sci Adv. 2020;6. \ndoi:10.1126/sciadv.abb1430 \n43.  Lea AJ, Caldas IV, Garske KM, Echwa J, Gurven M, Handley C, et al. Adaptations to water \nstress and pastoralism in the Turkana of northwest Kenya. bioRxiv. 2023. \ndoi:10.1101/2023.01.17.524066 \n44.  Lea AJ, Waigwa C, Muhoya B, Lotukoi F, Peng J, Henry LP, et al. Socioeconomic status \neffects on health vary between rural and urban Turkana. Evol Med Public Health. 2021;9: \n406–419. \n45.  Lea AJ, Garcia A, Arevalo J, Ayroles JF, Buetow K, Cole SW, et al. Natural selection of \nimmune and metabolic genes associated with health in two lowland Bolivian populations. \nProc Natl Acad Sci U S A. 2023;120: e2207544120. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\n46.  Braun V, Clarke V. Using thematic analysis in psychology. Qual Res Psychol. 2006;3: \n77–101. \n47.  Benjamini Y, Hochberg Y. Controlling the false discovery rate: A practical and powerful \napproach to multiple testing. J R Stat Soc Series B Stat Methodol. 1995;57: 289–300. \n48.  Neudecker H. Michael Greenacre, Jörg Blasius (2006): Multiple correspondence analysis \nand related methods. Stat Pap (Berl). 2010;51: 1017–1018. \n49.  Lê S, Josse J, Husson F. FactoMineR: AnRPackage for Multivariate Analysis. J Stat Softw. \n2008;25. doi:10.18637/jss.v025.i01 \n50.  Novak NL, Allender S, Scarborough P, West D. The development and validation of an \nurbanicity scale in a multi-country study. BMC Public Health. 2012;12: 530. \n51.  Fleskes RE, Gilmore JK, Oubré LS, Ofunniyin AA, Cabana GS, Schurr TG. Immersive \nvideography of ancient DNA extraction for community engagement and educational \ninitiatives by the Anson Street African Burial Ground Project. Am J Biol Anthropol. \n2025;186: e25055. \n52.  Riedlinger M, Massarani L, Joubert M, Baram-Tsabari A, Entradas M, Metcalfe J. Telling \nstories in science communication: case studies of scholar-practitioner collaboration. J Sci \nCommun. 2019;18: N01. \n53.  Begay R, Archuleta S, Allison-Burbank JD, Begaye V, Howe L, Jim L, et al. Indigenous \ncommunity-based participatory research approach and safe return to in-person learning for \ndiné youths and families. Am J Public Health. 2024;114: S350–S353. \n54.  Ferreira MP, Gendron F. Community-Based Participatory Research with Traditional and \nIndigenous Communities of the Americas: Historical Context and Future Directions. In: \nDigitalCommons@WayneState [Internet]. 2011 [cited 2 Feb 2026]. Available: \nhttps://digitalcommons.wayne.edu/nfsfrp/2/ \n55.  Cueva M, Kuhnley R, Revels L, Schoenberg NE, Dignan M. Digital storytelling: a tool for \nhealth promotion and cancer awareness in rural Alaskan communities. Int J Circumpolar \nHealth. 2015;74: 28781. \n56.  Cambou D. The UNDRIP and the legal significance of the right of indigenous peoples to \nself-determination: a human rights approach with a multidimensional perspective. Int J Hum \nRights. 2019;23: 34–50. \n57.  Carroll SR, Garba I, Figueroa-Rodríguez OL, Holbrook J, Lovett R, Materechera S, et al. \nThe CARE principles for indigenous data governance. Data Sci J. 2020;19. \ndoi:10.5334/dsj-2020-043 \n58.  Beaton A. Te Mata Ira: Guidelines for Genomic Research with Māori.: Te Mata Ira: \nGuidelines for Genomic Research with Māori. Hamilton, New Zealand: Māori and \nIndigenous Governance Centre, University of Waikato; 2016 Oct. Available: \nhttps://researcharchive.wintec.ac.nz/id/eprint/5640/ \n59.  Friedrich MJ. Ethical guidelines for genomic research in Africa. JAMA. 2018;319: 2371. \n60.  Kovach M. Indigenous methodologies. 2nd ed. Toronto, ON, Canada: University of Toronto \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint \n\nPress; 2021. \n61.  Dawson E. Equity, exclusion and everyday science learning. Abingdon, Oxon ; New York, \nNY : Routledge, 2019.: Routledge; 2019. \n62.  Cochran GL, Cochran HC, Ernst ME. Research and scholarly methods: Mitigating \ninformation bias. J Am Coll Clin Pharm. 2025. doi:10.1002/jac5.70054 \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted March 19, 2026. ; https://doi.org/10.64898/2026.03.17.711941doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}