Introduction

Homegardens are traditional agroecosystems found across tropical and subtropical regions, characterized by their structural complexity, high species diversity, and spatial integration within household compounds. These systems, often overlooked in global agricultural assessments, represent multifunctional land-use units that support a wide range of provisioning, regulating, and cultural ecosystem services (Kumar and Nair, 2004; Galluzzi et al., 2010). Composed of perennial and annual species arranged in multilayered configurations, homegardens contribute significantly to household food security, nutrition, health, and income (Torquebiau, 1992; Marsh, 1998; Soemarwoto, 1987). In certain contexts, they can provide up to 44% of caloric intake and 32% of protein needs, as well as essential micronutrients such as iodine and iron (Galhena et al., 2013; Kehlenbeck et al., 2007).

Beyond their nutritional value, homegardens play a key role in biodiversity conservation, acting as refugia for cultivated and wild species, including landraces and traditional varieties that are often absent from larger-scale production systems (Trinh et al., 2003; Galluzzi et al., 2010). Their species richness and floristic composition frequently rival those of natural ecosystems, and their management is embedded within local ecological knowledge and cultural practices (Kumar and Nair, 2004). Homegardens also contribute to microclimatic regulation, carbon storage—with reported carbon stocks comparable to secondary forests (Singh and Sohoo, 2021)—and erosion control, thereby enhancing the resilience of rural livelihoods to environmental and economic shocks (Pandey et al., 2006; Lakshmi et al., 2021).

Despite their ecological and socio-economic importance, systematic analyses of species composition in homegardens remain scarce. Most studies are site-specific and qualitative, with limited harmonization of botanical data across regions. This gap hinders our ability to generalize findings and to integrate homegardens into broader strategies for biodiversity conservation and sustainable agricultural development (Galluzzi et al., 2010; Peyre et al., 2006). In particular, there is a lack of large-scale, quantitative syntheses on the taxonomic diversity and spatial distribution of cultivated species in homegarden systems.

This study aims to address this gap by conducting a preliminary quantitative assessment of species composition in existing homegarden datasets across diverse geographic regions. By synthesizing and visualizing available data on cultivated species, this work provides a foundation for more advanced analyses of ecological functioning, including functional diversity, resilience, and ecosystem service provision in smallholder agroecosystems.

Methods

Literature search and eligibility criteria

We conducted a systematic review following established guidelines for evidence synthesis in ecology. The literature search was performed on June 2, 2025, using the Web of Science Core Collection. The search equation was designed to capture studies examining plant diversity in homegardens and was defined as:

(“home garden” OR homegarden OR homestead* OR “household garden*”) AND (species OR “species richness” OR “plant diversity” OR “biodiversity” OR “floristic composition”)**.

This query returned a total of 2,437 records. To supplement this corpus, we also manually screened the reference lists of relevant reviews for additional eligible studies not indexed in Web of Science.

Studies were included if they met the following criteria: (i) the study system was explicitly focused on homegardens; (ii) the data were derived from multiple homegardens (i.e., not single-case studies); and (iii) plant species were recorded through field sampling or botanical inventories. We excluded conceptual papers, reviews, and studies not involving direct species observations.

Screening and inclusion process

Title and abstract screening was conducted using the semi-automated platform Abstrackr, which allows for active learning based prioritization. Abstracts were ranked by their predicted probability of inclusion. Screening proceeded in descending order of predicted relevance until we reached a saturation threshold—defined as 100 consecutive abstracts without a single inclusion—after which the process was halted. This occurred at 1,540 screened abstracts. The final included set comprised 300 studies.

Data extraction and standardization

For each included study, we manually extracted the following information from the full text and supplementary materials: study title, country (and region if available), year of publication, total number of plant species recorded, number of homegardens sampled, and any reported diversity indices or explanatory variables. All geographic names were standardized using the countrycode package in R.

Species richness was defined as the total number of plant species observed across the sampled homegardens. Explanatory variables were also coded when studies explicitly linked diversity patterns to land use, management practices, landscape characteristics, or environmental gradients.

Statistical analyses and visualization

All statistical analyses were performed in R version 4.4.0. Geographic distribution of studies was mapped using rnaturalearth and sf. Choropleth maps were produced to visualize the number of studies per country as well as average species richness per country. Richness levels were grouped into four classes based on quartiles to enhance interpretability.

Descriptive summaries were generated for diversity metrics and explanatory variables. We also analyzed temporal trends in publication frequency and taxonomic focus. Figures were created using ggplot2, and all code is available in the project repository

Results

Number and date of publication

Our systematic review includes 300 studies analysing species composition in homegarden, published between 1990 and 2025. Early publications (1990–2004) were sparse, with fewer than five studies annually. From 2007 onwards, publication frequency increased steadily, reaching 10–13 studies per year between 2009 and 2016. A marked surge occurred from 2017, peaking at 27 studies in 2024. The period 2018–2024 alone accounts for over half of all included studies, reflecting heightened global attention to homegardens as key agroecological systems.

The number of homegardens analyzed per study varies widely, ranging from none reported to a maximum of 839. The median number of homegardens sampled is 73.5, with a mean of 109.8, indicating a generally substantial sampling effort. The interquartile range spans from 31 to 133 homegardens. These figures remain provisional as approximately 208 studies lack this information, pending further data characterization.

Global distribution of studies

The 300 studies included in this review span 55 countries, with pronounced geographic biases. Ethiopia leads with 50 studies, followed by India (35) and Mexico (27). Other countries with substantial representation include Bangladesh (18), Brazil and Indonesia (13 each), and South Africa (12). Together, these eight countries account for more than half of all included studies, underscoring the focus on homegardens in tropical and subtropical regions. Lower representation is observed in temperate regions, with most countries contributing fewer than five studies. This geographic distribution highlights existing research gaps and opportunities for future studies in underrepresented region

Species richness and diversity patterns

Reported species richness per study ranges from 6 to 1166 species (median = 127; mean = 183), while family richness ranges from 14 to 302 families (median = 46; mean = 56). Data are missing for 150 and 210 studies for species and family richness, respectively. These values summarize paper-level results; further analysis is needed at the individual homegarden scale.

Mean species richness in homegardens varied markedly across countries, with robust evidence from countries featuring multiple studies. India (n=35), South Africa (n=12), China (n=10), and Sri Lanka (n=6) consistently showed high richness (>200 species). Countries with few studies but very high richness estimates, such as Ireland (n=5) and Austria (n=2), require cautious interpretation. Moderate richness (150–225 species) was observed in Indonesia (n=13), the US (n=2), Nepal (n=4), and Argentina (n=5). Lower richness (<150 species) predominated in several African and Latin American countries with varying study counts. These patterns highlight clear geographic differences in homegarden biodiversity, while emphasizing the need to consider study effort to avoid biased conclusions.

Biodiversity in homegardens was quantified using a variety of metrics, most commonly species richness (n = 78), followed by Shannon diversity (n = 51) and measures of evenness (n = 29). Simpson-based indices were less frequently used (n = 22), while functional and multi-scale diversity metrics—such as functional diversity, alpha, beta, and gamma diversity—were rarely reported (each n ≤ 5). This indicates a predominant reliance on taxonomic richness and abundance-weighted indices, with limited integration of functional or spatial dimensions in biodiversity assessments.

Experimental design

Data collection methods varied across studies, with interviews (n = 120) and surveys (n = 102) representing the most common approaches. Fieldwork (n = 35), sampling (n = 31), and questionnaires (n = 30) were also frequently employed, reflecting a strong emphasis on direct, empirical data gathering. Less common methods included inventories (n = 29), case studies (n = 6), and direct observations (n = 6). Experimental designs (n = 3), monitoring (n = 2), and remote sensing (n = 1) were rarely used, highlighting limited application of long-term or technology-driven data collection in this research domain. These patterns suggest a predominance of qualitative and survey-based approaches in homegarden biodiversity studies.

Literature