Modern zoos play a multifaceted role, balancing entertainment with conservation, education, and research. Their *in situ* conservation contributions, primarily funded by visitor revenue, are substantial. A common belief is that large vertebrates, particularly mammals, are crucial for attracting visitors. However, these animals are expensive to maintain, challenging to breed, and raise ethical concerns. The conservation movement encourages a shift towards smaller species, native species, and threatened species, potentially leading to specialized collections. This shift, however, might negatively impact visitor attendance and thus *in situ* funding. The existing research on the relationship between zoo collection composition and attendance is limited in scope and often doesn't account for the indirect effects of various factors. This study aims to address this gap by developing a comprehensive model to understand the complex interplay between collection composition, socioeconomic variables, visitor attendance, and *in situ* conservation contributions, informing more effective collection planning and policy.

Previous studies have explored the factors influencing zoo visitor attendance, focusing primarily on the direct effects of variables like species composition and socioeconomic factors. These studies are often limited in scope, considering specific regions or species groups. While the “flagship approach,” utilizing popular large vertebrates to attract visitors and raise funds for *in situ* conservation has been discussed, a global, comprehensive analysis linking collection composition to both attendance and *in situ* contributions is lacking. Some research indicates a positive correlation between popular species (large vertebrates) and attendance, but the balance between visitor preferences and conservation priorities remains a challenge. This paper aims to expand upon this limited research by examining a global dataset and using structural equation modeling to capture the complexities of direct and indirect effects of variables.

This study employed structural equation modeling (SEM) to analyze the relationships between zoo collection composition, socioeconomic factors, visitor attendance, and *in situ* conservation contributions. Data were collected from multiple sources: Species360 for vertebrate collection composition (458 zoos worldwide), the International Zoo Yearbook (IZY) for attendance data (2015), and the Association of Zoos and Aquariums (AZA) for *in situ* project contribution reports (119 zoos). The variables included total number of animals, species richness (total and mammals), mean species body mass, collection dissimilarity, zoo area, proportion of threatened species, GDP, national population size, and population within a 10km radius. Two separate SEM models were developed: an attendance model (n=458) and an *in situ* model (n=119). The *in situ* model incorporated the attendance model's findings to leverage the larger sample size of the attendance model. The researchers used various statistical methods, such as calculating the mean abundance-weighted species body mass and employing model-data fit procedures to refine the SEM models using AICc values. Bivariate relationships between key variables were also visualized. Additionally, alternative models based on species presence-absence rather than abundance were developed for comparison. The analyses were performed using R statistical software.

The results demonstrate a positive correlation between high attendance and *in situ* conservation contributions. Zoos with high attendance, larger areas, and a higher proportion of threatened species tend to contribute more to *in situ* projects. However, collection composition variables (total number of animals, species richness, mammal species richness, dissimilarity, and species body mass) were stronger predictors of attendance than socioeconomic variables. The total number of animals showed the strongest positive direct effect on attendance, followed by mean species body mass. Mammal species richness had a positive direct and indirect effect on attendance. Species richness displayed a positive indirect effect mediated by the total number of animals but had a small, negative direct effect. There is a trade-off between the number of animals and body mass; high body mass negatively impacts the total number of animals and collection dissimilarity. While mean species body mass positively affected attendance, this direct effect was weakened when considering its indirect negative effects on the number of animals and dissimilarity. The proportion of threatened species did not directly or indirectly affect attendance, which contrasts with some earlier findings. The *in situ* model showed that attendance is the most significant predictor of *in situ* contributions. The proportion of threatened species was positively correlated with *in situ* contributions but not with attendance. The study highlights multiple pathways influencing attendance and *in situ* contributions; for instance, a zoo could increase attendance and contributions by either focusing on large-bodied animals or by focusing on many small, unique species.

The findings challenge the exclusive reliance on large vertebrates as a strategy to maximize visitor attendance and *in situ* conservation funding. While the flagship approach using large, popular animals is effective, alternative strategies focusing on high species richness, many smaller animals, and unique collections are equally viable. The positive correlation between attendance and *in situ* contributions underlines the importance of visitor engagement for conservation efforts. The lack of a direct link between the proportion of threatened species and attendance suggests that conservation-focused collections might not necessarily translate to higher visitor numbers. However, the positive correlation between the proportion of threatened species and *in situ* contributions emphasizes the importance of zoos' role in the conservation of threatened species. The trade-offs identified between various collection composition variables highlight the complexity of collection planning and the need for a system-wide approach that considers multiple direct and indirect effects.

This study provides novel insights into the complex relationships between zoo collection management, visitor attendance, and *in situ* conservation. The findings support the continued use of the flagship approach but also suggest alternative strategies for increasing attendance and conservation contributions. Future research could explore the relative cost-effectiveness of different collection strategies, investigate the specific types of *in situ* contributions made by zoos, and analyze the role of public perceptions and educational programs in driving conservation engagement.

The study relies on self-reported data from zoos, which may be subject to biases or inconsistencies in data collection and reporting. The *in situ* model had a smaller sample size compared to the attendance model, potentially limiting the generalizability of the findings. The use of attendance as a proxy for income to fund *in situ* activities might not fully capture the economic complexity of zoo operations. Finally, cultural variations in visitor preferences were not explicitly considered in the model.