Long-term transplant success is limited by allograft rejection, a complex process traditionally studied on an organ-specific basis. To establish a unified framework beyond organ-specific studies, we performed a network-based systems biology analysis of transcriptomic data from 672 liver, kidney, and heart transplant biopsies to identify a conserved, pan-organ molecular framework of rejection. By constructing and comparing organ-specific gene co-expression networks, we identified a consensus, six-module immune cascade that captures the hierarchical nature of the alloimmune response. In addition, we also uncovered a highly conserved 24-gene cell cycle signature consistently upregulated in rejecting allografts, implicating cellular proliferation as a core feature of rejection pathology. From this framework, we derived a 172-gene immune signature and applied machine learning models to assess its predictive performance, achieving accuracy comparable to established benchmarks. We further refined this to a minimal, high-performance 20-gene immune signature (AUC > 0.96). Both the immune and cell cycle signatures demonstrated robust, pan-organ utility when independently validated in a lung transplant cohort (n = 243). Collectively, these findings define a pan-organ molecular framework for rejection and highlight cell cycle dysregulation as a conserved hallmark, offering a foundation for standardized, cross-organ diagnostic platforms to improve allograft surveillance and patient outcomes.
Over the past two decades, genetic testing has undergone major shifts in its accessibility and in its nature. Historically, it primarily involved analysis of single genes selected on the basis of symptoms or family history, and was available only to a few. Now, options range from diagnostic clinical genomic tests, to broader screens offered to 'healthy' populations, to direct-to-consumer tests offering to explore ancestry. As genetic testing becomes an increasingly 'everyday' encounter, we sought to explore how the topics of genetics (and genomics) were considered in the Mass Observation Project, an archive of writing by 'ordinary' people about everyday life in Britain. 55% of the 147 respondents had personal experience of genetic testing or knew someone who had, typically to explore ancestry. Responses often gave the sense of genetic testing as a powerful tool in healthcare with results that were fairly definitive. Genomic testing was typically written about as an amplification of genetic testing, generating information of similar solidity. Writers threaded together personal experiences with insights drawn from a wide variety of media, often quite old, in outlining their ideas. While many positioned genetics as outside their remit, respondents engaged in depth with the opportunities and challenges raised, advocating for ethical/societal considerations to form a key part of decision-making regarding genetics. Our analysis shows that people without prior experience of clinical genetic testing may yet have a wealth of experiences and exposures sculpting their expectations as to what testing stands to bring. Consent conversations may benefit from exploring these.
Karyotype analysis is a core component of medical genetics education, yet many students find it challenging, particularly in large classes where opportunities for individualized feedback are limited. Digital tools may help address these constraints, but evidence on their use across different class sizes remains limited. This study examined the use of an interactive digital tool to support learning in karyotype analysis within undergraduate medical education. The study involved 252 first-year medical students who used an interactive digital tool integrated into the medical genetics curriculum and delivered asynchronously through the university's virtual learning platform. Students completed structured online activities requiring the construction and interpretation of karyograms from real metaphase images. Data were collected using pre- and post-activity surveys assessing self-reported knowledge, diagnostic confidence, and perceptions of learning. Descriptive statistics and comparative analyses were used to examine changes over time and to explore differences between small and large teaching cohorts. Following the intervention, students reported improved clarity of key genetic concepts and increased confidence in identifying chromosomal abnormalities compared with pre-activity measures. Most participants also perceived the tool as useful for supporting their learning and reported high levels of engagement with the activity. Similar patterns were observed in both small and large cohorts, suggesting that the digital tool functioned comparably across different class sizes. The findings indicate that interactive digital tools can successfully support the acquisition of complex scientific domains, particularly in contexts characterized by large enrollments and limited access to laboratory-based instruction. Although conducted within medical genetics education, the study has broader relevance for higher education settings seeking scalable approaches to teaching conceptually demanding content. The results may inform educational policy and curriculum decisions related to the integration of digital resources in health professions education. Not applicable.
As larger genomic data sets become available for wild study populations, the need for flexible and efficient methods to estimate and predict quantitative genetic parameters, such as the adaptive potential and measures for genetic change, increases. Even though animal and plant breeders, as well as the field of human genomics, have produced a wealth of methods, wild study systems often face challenges due to larger effective population sizes, environmental heterogeneity and higher spatio-temporal variation. Existing approaches either rely on two-step procedures, where residuals from a pre-fitted model are used as the response in a second analysis, or can become computationally inefficient as model complexity and data size increase. We therefore adapt methods from animal breeding to account for the complexity typically present in wild animal populations. The core idea is to approximate breeding values as a linear combination of principal components (PCs), where the PC effects are shrunk with Bayesian ridge regression. The result is a computationally efficient and scalable approach, denoted Bayesian principal component ridge regression (BPCRR). A case-study for a Norwegian house sparrow meta-population, as well as simulations, illustrate that the method efficiently estimates the additive genetic variance and accurately predicts breeding values. In order to assess whether BPCRR predicts informative breeding values, we also apply BPCRR to track micro-evolutionary change across time and space in the house sparrow system. To make the method accessible, we provide coded examples and data.
暂无摘要（点击查看详情）
Understanding how landscape features influence gene flow in natural populations is a central goal of landscape genetics. In this study, we evaluated the status of multiple populations of Camellia chekiangoleosa-a provincially protected plant in Jiangxi Province with significant ecological and economic value-in the Laohunao Nature Reserve. We then performed an analysis of their genetic diversity and spatial genetic structure to inform the development of science-based conservation strategies. The findings indicated that the genetic diversity of C. chekiangoleosa was on par with that of the majority of other species within the genus. Sub-populations exhibit a certain degree of genetic differentiation (F ST=0.065). The application of STRUCTURE and principal component analysis (PCA) unveiled a distinct pattern of geographical clustering across the 9 sub-populations. Multiple matrix regression with randomization (MMRR) revealed that both environmental isolation and geographical isolation have wielded a pronounced impact on genetic distance. The MEMGENE analysis further discerned spatial genetic structure, which was influenced by the ridges and valleys that demarcate Mount Ruozhushan (RZS), Mount Zhushan (ZS), and Mount Shitoushan (STS). Mountain ridges exerted a more pronounced influence on genetic differentiation compared to valleys. In summary, this study underscores that topographic features play a pivotal role in shaping the spatial genetic structure of C. chekiangoleosa.
Ageing-related diseases (ARDs) display diverse phenotypes yet share an age-dependent rise in incidence, suggesting mechanistic links with ageing processes. We examined whether ageing-related genes differ systematically from genes associated with multiple ARD clusters. Across 57 ARDs from UK Biobank, network analyses showed that ageing-related genes, although rarely ARD-associated, lie significantly closer to many ARDs through greater-than-chance proximity in protein-protein interaction (PPI) and KEGG networks. Consistent with this network overlap, ageing-related genes tend to occupy intermediate tiers in KEGG signalling cascades and exhibit strong coexpression with disease-associated genes as well as low tissue specificity, supporting that they play regulatory roles across multiple tissues. In contrast, genes associated with multiple ARDs are enriched for immune disorders and tend to have fewer ARD-associated neighbors in PPI networks. Accordingly, genes associated with multiple ARDs also tend to be located in terminal branches of KEGG pathways and show high tissue specificity coupled with weak coexpression with other ARD genes. Lastly, machine learning integration based on gene-disease network topology identified candidate ageing-related genes enriched for intracellular signal transduction and programmed cell death. Altogether, this work reveals two genetic architectures of multi-ARD influence: a cross-tissue regulatory mechanism enriched in ageing-related genes, and a tissue-specific, immune-driven mechanism among pleiotropic disease-related genes.
We discuss here physics and evolution of globular proteins with a focus on the basic units of their structure and function, closed loops and elementary functional loops. A starting point of the journey here is a prebiotic evolution, in which short linear peptides and corresponding RNA duplexes with traits determined by demands to survive and to move on in the harshness of the Origin of Life emerged first. Next, we follow the fate of ring-like peptides, which apparently were the "Dayhoff fragments" that passed through abiogenic transition, formed first functional domains followed by their inclusion in multidomain structures, formation of complexes, assemblies, and molecular machines in later stages. I argue that physics, specifically polymer nature of protein chains, not only served as a determinant of the basic units of proteins-closed loops but also could play a decisive role in determining the size of another important structural unit-domain-based on the polymer nature of nucleic acids that governed the optimal ring closure size of DNA molecules. The protein closed loops served as scaffold for carrying elementary functions-descendants of ancient ring-like peptides, combinations of which provided the modern protein function universe. Another alliance between physics and evolution discussed here is the allosteric regulation of protein function, which is based on the structural dynamics underlying the allosteric signal transduction and its regulatory role. While the physics drive the structure-based conserved patterns of allosteric signalling determined by the folds, the evolution brings in a sequence diversity, allowing to alternate the allosteric communication and to make it archetypal for distinct functional (super)families using the same fold as the structural platform. Considering only few above aspects, I show that important lessons from the hand-in-hand walk of physics and evolution already helped to achieve the current state-of-the-art in understanding of protein structure and function. I conclude, however, that there is still even more to learn from Nature about long and remaining mainly mysterious 3.5 billion years endeavor.
Memories can last a lifetime, and how this is achieved remains an unanswered challenge. Most current thinking sees molecular traces of memories (engrams) as sets of synaptic proteins facilitating neuronal co-firing and co-wiring. However, most proteins turn over in months or less. Another challenge is how fibroblasts remember their cell fate for decades, and an emerging model sees functionally related genes co-firing in clusters (called transcription factories and condensates) that make RNAs specifying cell fate. As clustering is driven by entropic forces acting throughout time, the first cells may have possessed this memory system, and Nature could have exploited it to store engrams when nervous systems evolved. Then, transcription creates the naïve neuronal substrate and defines which cells are included in co-wiring and co-firing circuits, before progressive cell differentiation consolidates long-term memories. I speculate that transcription plays another central role. For every nucleotide added to a nascent RNA, transcription generates a pyrophosphate-a chelating agent that sequesters the calcium ions that can modify action-potential spike-trains. In other words, the same nano-wired DNA computer that specifies cell fate could store and manipulate our memories.
Urban trees and their phyllosphere-associated microbiota constitute a promising nature-based solution for mitigating urban air pollution. In this study, we characterized the taxonomic composition, diversity patterns, and functional potential of bacterial communities inhabiting the phyllosphere of Mangifera indica in two urban sites of Medellín, Colombia, with contrasting pollution levels and across two time points, analyzing a total of 12 samples. We integrated 16S rRNA gene amplicon sequencing, performed on the Illumina MiSeq platform, with shotgun metagenomic sequencing generated on the Illumina NovaSeq 6000 platform to assess community structure and the presence of genes involved in the degradation of airborne organic pollutants. Bacterial assemblages were dominated by Pseudomonadota (Proteobacteria), Actinomycetota, and Bacteroidota, with genera such as Methylobacterium, Pseudomonas, and Serratia consistently prevalent. Alpha diversity was higher in the highly polluted downtown, while beta diversity was shaped primarily by temporal variation. Functional annotation of metagenome-assembled genomes (MAGs) uncovered genes encoding complete aromatic hydrocarbon degradation pathways, including naphthalene, toluene, xylenes, and benzoate. Both ortho- and meta-cleavage routes for catechol degradation were detected, with temporal shifts in pathway dominance linked to changes in the abundance of key degraders taxa. These results reflect genetic potential for xenobiotic degradation within the M. indica phyllosphere microbiota, modulated by environmental conditions. Our findings highlight the ecological role of phyllosphere bacteria as contributors of inferred functional capacity relevant to atmospheric bioremediation and supports their integration into microbiome-informed green infrastructure strategies.
This study analyses relationships between milk yield, milk composition, and fertility in Murciano-Granadina dairy goats to test whether these associations reflect biological pathways underlying a production-fertility trade-off. Linear and non-linear patterns were examined using canonical discriminant analysis (CDA), regularized canonical correlation analysis (rCCA), and CHAID decision trees. The dataset included 32,693 artificial inseminations from 21,757 does and 29,390 milk records from 21,541 does; fertility was assessed using three indicators accounting for insemination timing and semen type. Mean fertility differed by semen preservation method, with fresh/chilled semen showing 18-20 percentage points higher fertility than frozen/thawed semen. Milk yield showed wide variability (8.7-6704.6 kg; mean 686.4 kg), whereas milk composition was relatively stable. After multicollinearity control (VIF > 178 to <5), CDA and rCCA revealed a low-dimensional structure dominated by a first canonical function, explaining 80.8-88.7% of discriminant variance and 97.3% of shared covariance. This dominant axis reflected an energy balance and metabolic partitioning pathway, with higher milk yield and solids concentration associated with reduced fertility; very high production (>2021 kg) consistently coincided with lower fertility, supporting a production-reproduction trade-off. A secondary axis represented an endocrine and lactation-timing pathway linking fertility to temporal variation in milk composition, while semen type constituted an autonomous semen-related pathway with the strongest standardized canonical coefficients. CHAID analysis identified non-linear relationships and biologically meaningful thresholds in milk traits, achieving ~44% classification accuracy and up to 87% reliability for high fertility. Optimal fertility was associated with intermediate ranges of milk yield and composition (protein 3.64-6.98%, fat 6.0-7.9%, dry matter 14-25%, lactose 5.6-6.5%, SCC < 5200 × 10³ cells/mL). Overall, milk yield and composition showed statistically robust but moderate associations with fertility, consistent with reproductive biology multifactorial nature of and gaining relevance when interpreted jointly with metabolic, endocrine, and semen-related factors.
Multimorbidity, also known as multiple long-term conditions, is a major public health concern. Internalising and CardioMetabolic MultiMorbidity (ICM-MM) is a common form of mental-physical health multimorbidity, yet its genetic predisposition is largely unknown. We examined the polygenic nature of ICM-MM by assessing single trait-specific polygenic risk scores (PRSTRAIT) and whether combining them could increase the proportion of variance in liability to ICM-MM explained by genetic variation. We developed PRSTRAIT using PRS-CS and summary statistics from the largest trait-specific GWAS excluding UK Biobank (UKB). We evaluated PRSTRAIT on ICM-MM risk in 206 452 UKB participants (n = 39 311 (19.0%) with ICM-MM) using logistic regression adjusted for gender and 10 genetic principal components, defining ICM-MM as lifetime occurrence of: ≥1 internalising (depression, anxiety, somatoform disorder) traits AND ≥ 1 cardiometabolic traits (type 2 diabetes, obesity, hypertension, dyslipidemia, chronic kidney disease). We used elastic net regression in a 50% training sample to generate ICM-MM-PRSTRAIT: a weighted combination of PRSTRAIT targeting ICM-MM. The strongest associations were between ICM-MM and PRSTRAIT for depression and type 2 diabetes-both odds ratios (OR) 1.18, [95% confidence interval (CI) 1.17-1.20] per standard deviation increase in PRSTRAIT. ICM-MM-PRSTRAIT retained five PRSTRAIT, with stronger associations (OR = 1.31, [95%CI 1.29-1.34]) than any PRSTRAIT in the testing sample. Combining several PRS explains more variance in ICM-MM liability than single-trait PRSs alone. ICM-MM-PRSTRAIT is a measure of genetic risk that could be used to examine premorbid stages of ICM-MM in external and youth cohorts, supporting awareness of earlier presentation and potentially avoidance or intervention.
Several MLLT10-associated single-nucleotide polymorphisms (SNPs) have been identified by genome-wide association studies (GWASs) as germline risk variants for meningioma in predominantly European cohorts, but their relevance in Koreans remains uncertain. We investigated these MLLT10 risk SNPs in Korean meningiomas, assessing differences across two time cohorts and comparing allele frequencies with those observed in other populations. Three MLLT10 SNPs (rs12770228, rs11012732, and rs1243180) were examined in 143 meningiomas from patients aged ≤50 years, comprising 62 fresh-frozen tissues collected during 1999-2003 (Period 1) and 81 formalin-fixed paraffin-embedded tissues from 2006-2023 (Period 2). Three SNPs were detected in 9 of 143 meningiomas (6.3%). While the differences did not reach statistical significance (p > 0.05), minor allele frequencies of all three SNPs were reduced two- to four-fold in Period 2 compared with Period 1. The observed frequencies were similar to those reported in Japanese cohorts but substantially lower than the ≥30% reported in European populations. Despite the limitation of using tumour-derived DNA to assess germline variants, our findings consistently showed that MLLT10 risk SNPs occur at very low frequencies in Koreans, similar to Japanese data and in contrast to Europeans. These results highlight the population-specific nature of MLLT10 variants and underscore the need for large-scale Asian studies for risk SNP analysis in meningiomas.
Precision oncology in urology increasingly depends on integrating heterogeneous data, including multiparametric imaging, histopathology, genomics, and clinical variables. Multimodal artificial intelligence (AI) offers a unified framework to manage this complexity, supporting refined risk stratification, personalized treatment decisions, and informed patient counseling. This narrative review examines applications of multimodal AI in prostate, bladder, and kidney cancers. Beyond listing individual tools, we emphasize how synergistic data fusion enhances the validation of diagnostic and prognostic performance. Clinical advances include more accurate tumor delineation on multiparametric MRI and predictive modeling of functional outcomes after surgery, underscoring the translational potential of these systems. However, major barriers hinder clinical adoption. Prospective validation remains scarce, data harmonization across institutions are limited, and the opaque nature of many algorithms fuels skepticism among clinicians. These factors collectively restrict the integration of multimodal AI into routine clinical practice. Closing this gap requires standardized data curation, development of interpretable and transparent models, and the design of collaborative human-AI workflows. Ultimately, successful translation will depend not only on technical progress but also on redefining trust and expertise in urologic oncology, ensuring that algorithmic insights are meaningfully aligned with bedside decision-making.
Convergent evolution, the repeated evolution of similar phenotypes, is widespread in nature, but there are few studies investigating the genetic mechanisms of convergence across wide evolutionary timescales. The extent to which the same genetic mechanisms contribute to convergent evolution could reveal whether the pathway towards these fitness optima is flexible or constrained to follow a particular route, informing us about the predictability of evolution. Wing color pattern mimicry in Lepidoptera is a well-known example of convergent evolution, but as studies are restricted to a few closely related species, it is difficult to make general inferences about the predictability of evolution in this system. Here we study convergent evolution in multiple mimetic neotropical lepidopteran lineages that diverged between ~1 and 120 Mya, including seven species of Ithomiini and Heliconius butterflies and a day-flying Chetone moth. Across butterfly lineages that diverged up to ~30 Mya, the genetic variants most strongly associated with convergent color pattern switches are located in similar noncoding regions near the genes ivory and optix. In the more distantly related moth species, color pattern variation is associated with a ~1 Mb inversion which also contains ivory, closely mirroring the supergene architecture of the co-mimetic butterfly Heliconius numata. In contrast to previous studies on Heliconius butterflies, we find limited evidence that convergence among closely related Ithomiini species results from alleles shared by hybridization. Repeated parallel evolution of regulatory switches via reuse of the same two genes suggests that convergent color pattern evolution is highly constrained and predictable even across large evolutionary timescales. Such constraints may have facilitated diverse taxa joining this species-rich mimicry ring.
The protease TMPRSS2 facilitates coronavirus infections, yet its mechanism of viral glycoprotein recognition remains unclear. Here we show that, following ACE2 engagement of the SARS-CoV-2 spike (S) inducing the early fusion intermediate conformation (E-FIC), TMPRSS2 cleaves the R815 S2' site and promotes fusogenic conformational changes leading to viral entry. We unveil TMPRSS2 recognition of S2', identify key residues modulating binding specificity and demonstrate that S2' site-directed broadly neutralizing antibodies target E-FIC and inhibit viral entry by blocking TMPRSS2 access. We computationally designed stabilized E-FIC as a vaccine candidate, overcoming the transient nature of this state. We describe a TMPRSS2-directed monoclonal antibody inhibiting several coronaviruses, including SARS-CoV-2 variants and protecting mice against SARS-CoV-2 challenge. These results outline the mechanistic role of TMPRSS2 and S2' site-directed antibodies in coronavirus entry.
Conventional transcriptomics analyzes a mixture of cells simultaneously either from tissue or single cells, for which the information on its in vivo or in situ position is lost. Although we can now perform cell type annotations for heterogenous samples using single-cell profiling, information on the spatially defined attributes that underpin the heterogeneous nature of the cells is lacking. Here, we describe a further optimized 10× Visium spatial transcriptomics protocol, a method that enables the deconvolution of heterogenous tissues while retaining its spatial information. The 10× Visium protocol allows the transcriptomic profiling of both fresh frozen and FFPE tissues while retaining their native spatial information. This protocol has wide potential applications to address the intricacies of the tumor microenvironment such as cellular interactions between the different tissue compartments and or with immune cells and elucidate how those interactions may contribute to tumorigenesis or disease progression. The 10× Visium spatial transcriptomics assay has been widely applied to study the spatial transcriptome of various tissue types across human and mouse samples such as lung, melanoma, and endometrial cancer to name a few. The entire protocol from tissue collection to final library preparation for fresh frozen samples takes about 5 days and for FFPE samples, about 3 days. Sequencing time ranges from 2 to 4 weeks and varies between different sequencing vendors.
Fatigue is a common and disabling non-motor symptom in Parkinson's disease (PD), significantly affecting patients' quality of life. However, it is often underdiagnosed due to its subjective nature and the lack of a clear definition, hindering the development of effective treatments. This study aims to investigate the prevalence of fatigue and its associations with sociodemographic factors, disease severity, levodopa equivalent daily dosage (LEDD) and motor, non-motor, and cognitive symptoms in an Italian cohort of patients with PD. An observational cross-sectional study was carried out in three Italian centers from January to May 2024. One hundred PD patients (H&Y ≤ 4) were assessed using validated tools: Parkinson Fatigue Scale (PFS), Fatigue Severity Scale, and Modified Fatigue Impact Scale. Motor and non-motor signs and symptoms, cognitive status, and LEDD were analyzed using non-parametric tests and Spearman's correlations. Fatigue prevalence was determined based on PFS score ≥ 3.09. Fatigue was present in 36% of patients, more prevalent in women and more severe in those with H&Y > 2. Fatigue correlated strongly with non-motor symptoms (MDS-UPDRS Part I; ρ > 0.6, p < 0.001) and moderately with motor complications (0.4 < ρ < 0.5, p < 0.001), but weakly with disease duration, LEDD and age (ρ < 0.3, 0.002 < p < 0.05). Significant intercorrelations among fatigue scales supported their ability to consistently measure the fatigue construct. Fatigue in PD is a multidimensional phenomenon influenced mainly by non-motor symptoms. Gender-specific differences and the association with disease progression underscore the need of comprehensive and integrated management strategies to address this challenging symptom.
Global efforts to mitigate anthropogenic pressures on biodiversity and ecosystems will often be realised through management at landscape-scales (i.e., in the range of 100s-1000 s km2). In consequence, we need to measure biodiversity responses at landscape-scales to ensure mitigations are effectively protecting and restoring ecosystems. Yet many countries currently lack monitoring programmes that can generate indicators of biodiversity at these scales. Localised monitoring (e.g., 1 km2) is often amalgamated into national-scale indicators, however, this leaves a substantial gap in the middle of this spatial gradient, limiting availability of information at decision-relevant scales. Here, using the United Kingdom as a case study, we explored the suitability of seven sources of biodiversity data which could be used to construct landscape-scale indicators. We surveyed 70, mostly UK-based, monitoring experts for their opinions on structured and unstructured in-person surveys, camera traps, eDNA, drones, passive acoustic recorders, and satellite remote sensing. We assessed data source utility to construct indicators reflecting Essential Biodiversity Variables, i.e., as holistic measures of taxa or ecosystems rather than assessments of individual management interventions. All seven data sources were deemed suitable, and experts expected developments in technology and infrastructure to greatly increase this potential over the next decade. However, there are technical, analytical, logistical and financial barriers to establishing monitoring networks that could yield the requisite data for landscape-scale indicators. Resolving these issues requires substantial research, policy commitment and investment, but landscape-scale indicators will be essential for the UK to undertake adaptive management and monitor nature recovery.
Visual camouflage evolution in animals depends on both light's interaction with its surroundings and the limits of what the natural observers of the camouflage can see. Changes in lighting - such as those caused by shifting weather - can quickly alter how animals and their surroundings appear due to the generation of shadows both cast onto objects and from self-shading. The extent and nature of these changes will depend on the three-dimensional (3D) structure of the environment. Despite the apparent effects of lighting on object and background appearance, the enormity of interactions and the diversity of animal camouflage methods and appearances pose challenges to investigating the combined effects of lighting and habitat structure on camouflage. Genetic algorithms and mathematical methods for generating animal patterns provide a potential solution for investigating camouflage's broad feature space by evolving artificial prey under different lighting conditions within different habitats. Here, an online artificial evolution experiment was used to examine the effect of lighting, both direct and diffuse, and habitat geometry on camouflage. Lighting and geometry changed the appearance of the evolved prey, and the predictive power of common measures of camouflage. Crucially, lighting condition systematically altered how contrasting the prey-targets' internal patterning was and interacted with habitat geometry to affect the evolved pattern shapes, colours, and countershading. Our work demonstrates the importance of considering the relative geometry and lighting of an environment when determining the function of animal colouration and the adaptive value of camouflage.