Crop diversity is an important factor in food and livelihood security, sustainable agriculture, and ecosystem services production [1
]. Crop diversity also serves as a hedge against risk by reducing farm-level vulnerability to climatic change or commodity market shocks [4
]. Erosion of crop diversity is of growing concern to global food systems security, the health and primary productivity of agricultural landscapes, and the long-term stability of socio-environmental systems [6
]. Enhancing crop diversity is now a core component of agri-environmental policies around the world, including multiple initiatives of the Food and Agriculture Organization of the United Nations [10
]. Despite this policy focus, many of the drivers of crop diversification are poorly understood [12
Studies often frame the drivers of crop diversification categorically as factors associated with broad farm types or cultivation strategies. A prime example is agricultural intensification, which is widely recognized as a key driver of diversity loss [14
]. Agricultural intensification tends to have homogenizing effects on landscapes that lead to specialized monocropping and erosion of biodiversity [16
]. Therefore, high-intensity, conventional agricultural systems are typically associated with lower crop diversity, while low-intensity, more traditional or agroecological systems are often associated with higher crop diversity. Within this framework, the categorical effects of agricultural intensification on crop diversity are clear. Less clear is how the partial (individual) effects of conventional inputs (e.g., chemical fertilizers, mechanization, irrigation) impact crop diversification.
There are at least two reasons why disaggregating ‘intensification’ and examining its partial effects on diversity is important. First, farm systems today often incorporate the characteristics of multiple farm types, blending attributes of traditional, agroecological, conventional, industrial, and other systems [19
]. In Mexico, for example, many low-input, traditional, smallholder farm systems nonetheless use high levels of chemical fertilizers [22
]. While understanding relationships between general farm system types, intensification levels, and crop diversity is informative, this level of analysis obscures the effects of specific inputs on diversity.
Second, the effects of intensification on crop diversity are poorly understood in the context of sustainable intensification. Increasingly, sustainable intensification is recognized as crucial to meeting future demands for food and minimizing environmental degradation [23
]. Although enhancing crop diversity is a key component of sustainable intensification [25
], it is conspicuously unclear how the mechanisms of intensification, whether conventional or sustainable, enhance crop diversity. In part, this is because high crop diversity is typically associated with low-intensity systems. While research on ‘scaling-up’ low-intensity practices to meet future demands for food is promising [28
], many conceptual and practical barriers remain [30
More immediate and feasible pathways to sustainable intensification are needed. Conventional input strategies have variable impacts on productivity and the environment. If properly managed, some conventional strategies can boost productivity without significant biodiversity loss [32
]. Though again, more research is needed to determine the effects of specific inputs on diversity loss or gain. Ultimately, a better understanding of input-diversification relationships is critical for transitioning to sustainable intensification—a transition that will inevitably require tradeoffs between conventional and agroecological approaches [33
Mexico presents an ideal case study of input-diversification relationships. Existing research has focused largely on maize genetic (landrace) diversity and the traditional intercropping systems (milpa) where much of this diversity is found [37
]. Beyond maize genetic diversity, less attention has been given to other forms of crop diversity, especially at higher taxonomic levels (e.g., species) and larger spatial scales (e.g., national level). Further, while several studies have examined the potential effects that changes in irrigation would have on crop production in Mexico—changes desperately needed to address growing regional- and national-level water scarcity [40
]—the potential effects of such changes on crop diversity are poorly understood.
To address these research gaps, this study seeks to answer two questions: (1) Does irrigation lead to higher or lower crop species diversity (hereafter, CSD) at regional and national levels in Mexico? (2) Are the effects of irrigation on CSD different in low-, medium-, and high-diversity regions? To answer these, we examine irrigation effects on crop species richness and evenness diversity using a quantile regression approach. We compare irrigation effects across conditional quantiles of low-, medium-, and high-diversity regions after adjusting for a range of socioeconomic, environmental, and farm characteristic factors. Findings are discussed in the contexts of water resources management and agricultural policy in Mexico, and the broader role of irrigation in sustainable intensification.
About 19% of municipality cropland received irrigation and about 28% received chemical fertilizers (Table 1
). About one-half (49%) of cropland was dedicated to maize cultivation, which broadly aligns with previous studies [61
]. On average, about 25% of farms cited the costs of inputs as a primary challenge, about 25% cited soil infertility, and more than one-half (54%) cited the challenge of commercializing or marketing crops. On average, about one-third (32%) of farms relied on mechanized labor and about 72% practiced subsistence agriculture. About 65% of municipalities had small (2–5 ha) or medium (5–15 ha) MFA (35 and 30%, respectively), and about 17%, 14%, and 4% had very small (0–2 ha), large (15–50 ha), or very large (>50 ha) MFA.
3.1. Model Results (OLS)
OLS models explained about 40% of the variances in both crop richness and evenness diversity, with adjusted r-squared values of 0.38 and 0.39, respectively (Table A1
and Table A2
). These values were similar to that of a previous national-level study on the determinants of CSD [47
]. The irrigation land ratio had the strongest positive effect on richness diversity (0.30), which was twice the effect of the next strongest positive predictor, location in the light semiarid (D3) climate region (0.15). The irrigation land ratio also had the strongest positive effect on evenness diversity (0.31), followed by subsistence agriculture (0.19), medium MFA (0.19), and location in the humid (B) region.
The strongest negative effects on richness diversity were location in the humid (B) region (−0.33), marginality level (−0.24), very small MFA (−0.17), mechanization challenges (−0.12), and the maize land ratio (−0.11). The strongest negative effects on evenness diversity were the maize land ratio (−0.50), locations in the semiarid dry (D1) and semiarid moderate (D2) regions (−0.31 and −0.18, respectively), and very large MFA (−0.17).
3.2. Model Results (Quantile)
The irritation land ratio also had the strongest positive effects on both richness and evenness diversity under quantile regression (Figure 2
a,b). Importantly, the positive effects on evenness diversity were up to five times larger in lower quantiles than in higher quantiles: effects in the 10th, 20th, and 25th quantiles of evenness diversity were 0.36, 0.50, and 0.35, respectively; while in the 75th, 90th, and 95th quantiles the effects were 0.18, 0.13, and 0.10, respectively (Figure 2
b). Additionally, the positive effects of medium MFA on richness diversity were more than three times as large in lower quantiles than in higher quantiles, which differed significantly from OLS results (i.e., where the 95% confidence intervals did not overlap).
The negative effects identified in OLS regression were mostly confirmed by quantile regression, except in the case of the maize land ratio effect on evenness diversity. Though maize continued to have the strongest effects, these varied by quantile, falling well below the lower boundary of the OLS confidence band (Figure 2
b). In contrast to the effects pattern of irrigation, the negative effects of the maize land ratio were about twice as large in higher quantiles of evenness diversity than in lower quantiles (−0.8 to −0.4, respectively). In other words, while the maize land ratio negatively predicted crop evenness diversity across all quantiles, the effects were twice as large in municipalities with higher, rather than lower evenness diversity. As in the case of the irrigation land ratio, these quantile distinctions were not detectable under standard OLS regression.
3.3. Forest Plot Comparisons
Side-by-side comparison of richness and evenness models showed irrigation effects were relatively consistent and stronger at lower, rather than higher quantiles (Figure 3
). In contrast, the negative effects of maize on richness and evenness diversity were less consistent, with strong negative effects on evenness diversity but weaker effects on richness diversity. In addition, while the negative effects of maize on richness were small across all quantiles (−0.06 to −0.11), the effects on evenness were large, ranging from −0.39 in lower quantiles to −0.77 in higher quantiles.
Aridity generally had negative effects on evenness diversity, though its effects on richness diversity varied widely across quantiles. These tended to be positive in the moderately dry regions (e.g., C2, D3, D2), negative in the most humid region (A), and small and insignificant in the most arid regions (e.g., D1 and E). Overall, analyses show that the effects of climate on crop diversity, once adjusted for other factors, were either small or statistically insignificant compared to the effects of the irrigation and maize land ratios.
3.4. Rainfed and Irrigated Crop Diversity
When stratified across climate regions, the richness of rainfed CSD varied little, with small and statistically significant differences in only a few regions (Figure 4
). The evenness of rainfed CSD varied even less, with no significant differences detected across climate regions. In contrast, irrigated CSD was generally higher than rainfed CSD, except in the most humid (A) region. The gap between rainfed and irrigated CSD tended to widen as aridity increased. The increase was also reflected in the irrigation land ratio—about 70% of cropland received irrigation in the two most arid regions (D1 and E), while less than 22% received irrigation in all other climate regions. The reverse trend was observed for maize, which was cultivated on less than 10% of cropland in the D1 and E regions, but on more than 40% in all other regions.
At the municipality level, crop species richness and evenness tended to be highest in Mexico’s northern regions, including the Baja Peninsula, northern Pacific coasts, and northern Central Tablelands (Figure 5
A,B). Species diversity tended to be lowest in the Southern Highlands region, northwestern mountains, and some areas in the Yucatán Peninsula. The share of cropland receiving irrigation was highest in municipalities of the northern border regions and in isolated patches of central and south-central Mexico (Figure 5
C). In contrast, the share of cropland cultivated with maize was highest in the Southern Highlands, Chiapas, the Yucatán Peninsula, and the northwestern and eastern mountains regions (Figure 5
4.1. Irrigation Enhances Crop Species Richness and Evenness Diversity
The main objective of this study was to determine how irrigation influences CSD at regional and national levels in Mexico. As a broad category of agricultural change, intensification generally leads to crop diversity loss, though the effects of individual inputs on diversity have been poorly understood.
Irrigation is a primary component of agricultural intensification, whether defined as a farm input [63
], a driver of increased productivity [64
], or both [65
]. Irrigation is also at the center of debates over the future of agricultural systems, water resources management, and sustainable development [66
]. Although the relationship between irrigation and crop diversity has received little direct attention, existing studies suggest two distinct types of effects.
First, irrigation leads to greater crop diversity when farmers take advantage of the broader range of crops that can be grown due to enhanced water availability [26
]. This pattern has been observed in farm-, landscape-, and regional-level studies in India [68
], Bolivia [69
], Nepal [70
], Bangladesh [71
], Ethiopia [4
], and several locations in Sub-Saharan Africa [73
]. In these cases, irrigation-led diversification often produces value-added crops that can be sold for a greater profit or more nutritionally diverse crops, which can enhance food and nutritional security [74
In contrast, irrigation leads to lower crop diversity (greater specialization) when farmers use the water to increase productivity (yield) but instead focus on a few water-intensive, high-yielding monocrops [75
]. Examples are found in several Asian counties where expanded access to irrigation disincentivized shifts toward alternative crops and instead increased cultivation of water-intensive rice [78
]. In Bangladesh, irrigation led to higher productivity but also to increased specialization in water-intensive wheat, which decreased overall crop diversity [76
]. Under both positive- and negative-effect scenarios, studies cite the contextual nature of farmer decisions about diversifying or specializing production, which are largely driven by farm-level responses to perceived opportunities and constraints [80
Only a few studies have assessed the relationships between irrigation and diversification at national levels. Studies from Slovakia [47
] and the United States [82
] found irrigation expansion led to greater CSD, while a study from China instead found it instead led to specialized monocropping [12
]. In general, national-level understanding of the relationship is insufficient, which has served as a barrier to effective policies [83
]. These policies increasingly view enhancing crop diversity as central to meeting sustainable development goals [26
This study found a positive relationship between irrigation and CSD in Mexico. At regional and national levels, as the share of cropland receiving irrigation increased, crop species richness and evenness increased. This relationship held after controlling for socioeconomic and environmental factors and multiple farm structural and functional characteristics.
In the context of previous research on relationships between irrigation and farmer decision making (above), our study shows that in the aggregate (municipality level), farms in Mexico employ irrigation more to diversify than to specialize crop species production. Regionally, the effects of irrigation on CSD are tied to the availability of existing irrigation infrastructure and other farm-level factors. Though beyond the scope of this study, previous work shows that regional differences in irrigation intensity (Figure 4
c) are strongly tied to existing socioeconomic inequalities and the influence of trade agreements [41
]. As explained below, these differences also manifest as regional differences in municipality-level CSD.
4.2. Irrigation Has Stronger Effects in Regions of Low Crop Species Diversity
The second objective of this study was to determine how irrigation influences CSD across conditional quantiles of low-, middle-, and high-diversity regions. We found that while most variable effects were small or statistically insignificant, those of irrigation were large and statistically significant across quantile ranges. Further, the positive effects of irrigation on species richness were almost twice as large in low-diversity quantiles compared with high-diversity quantiles. An even greater difference was observed with species evenness, as irrigation effects were five times larger in low-diversity compared with high-diversity quantiles.
Interestingly, these findings align with a recent study on the marginal effects of irrigation on maize and wheat yield in Mexico. The study found diminishing marginal returns on yield from increases in the irrigation land ratio in municipalities already receiving high levels of irrigation [22
]. The quantile effects on species diversity identified in this study, though distinct from the marginal effect on yield, suggest another form of diminishing returns from irrigation inputs. Together, both findings have implications for agricultural policy and water resources management in Mexico.
Recent studies also highlight the potential benefits of expanding irrigation access in southern Mexico. Southern agricultural regions are largely characterized by rainfed, maize-based cultivation, where crop water scarcity and low access to irrigation contribute to chronically low productivity [22
]. However, southern Mexico has the country’s largest reserves of replenishable freshwater resources [89
]. Southern regions also have among the highest rates of poverty and are home to many marginalized indigenous communities. Targeted expansion of sustainable irrigation infrastructure (i.e., soft-path approaches) in southern regions could contribute to numerous sustainable development objectives aligned with Mexico’s National Water Program priorities [40
Calls to expand irrigation in southern regions come with growing recognition that irrigation strategies in northern regions are unsustainable. The large-scale hydraulic infrastructure built during the 20th century (i.e., hard-path infrastructure) is outdated, and agricultural water-use efficiency has plummeted in the region [41
]. However, as this study illustrates, municipalities in these regions have among the highest levels of crop species diversity in the country—diversity that is strongly dependent on existing irrigation infrastructure. This pattern of dependency is similar to that of the United States, where the nation’s highest levels of crop species diversity are largely dependent on unsustainable irrigation practices in California [82
The strong effects of irrigation expansion on CSD observed in lower-quantile regions of southern Mexico adds support to calls for greater irrigation investment in this water-rich but irrigation-poor region, a condition expressed as agricultural-economic water scarcity [90
]. To be effective and sustainable, irrigation expansion in the region must be: (1) based on participatory approaches to integrated watershed management [91
], (2) carefully planned and targeted to priority regions and tailored to farm-level capacities and needs; (3) primarily limited to existing farmland [27
], and (4) focused on building small-scale, ‘soft-path’ infrastructure that preserves environmental flows [92
However, before changes in policy are made, a better understanding of the effects of irrigation on crop diversity at multiple taxonomic levels is needed. Specifically, understanding of the potential impacts of irrigation expansion on maize genetic diversity in Mexico is insufficient.
4.3. Crop Species Diversity and Scale: Important Distinctions
Importantly, our findings do not suggest that farm-level
crop diversity is necessarily lower in regions of low municipality-level diversity (e.g., Southern Highlands). The level of spatial aggregation is a key consideration for measuring crop diversity, as larger-scale measures often differ from farm-level measures [48
]. The level of spatial aggregation is especially relevant when assessing diversity across heterogeneous agricultural landscapes, where diversity measures are highly scale dependent [94
To illustrate this point, Figure 6
depicts a hypothetical model of two Mexican municipalities (A and B) with different crop compositions. When measured at the farm level (interior circles), crop richness and evenness diversity are higher in municipality B (see also Figure 1
). When measured at the municipality level (exterior circles), both richness and evenness diversity are higher in municipality A (more crop types and more even abundance of types).
We suspect a similar pattern exists in regions of Mexico where farm-level diversity and municipality-level diversity contrast. If confirmed, the pattern would be consistent with: (1) the high municipality-level species richness and evenness diversity in northern irrigated regions observed in this study and (2) the high farm-level species diversity observed in smallholder milpa systems of southern Mexico [53
]. Additional research on the gamma-, alpha-, and beta-diversity of crops in Mexico is needed to confirm this pattern [95
], as is additional research into the possible interaction effects of farm-level factors on CSD.
4.4. Other Limitations
The determinants and effects of diversification vary widely according to the taxonomic level of crops under study [18
]. In this study, we examined species
-level crop diversity, including the singular species of maize (Zea mays
L.). We did not consider the rich diversity of maize subspecies, varieties, and genetic (landrace) populations. Indeed, maize-based intercropping systems (milpa) are recognized as key reservoirs of in situ maize genetic diversity [98
]. The regions of high maize genetic diversity identified in previous research [99
] strongly correlate with the regions of high maize land ratios identified in Figure 4
d. At the crop species level, however, we found that the same regions tended to have lower municipality-level crop richness and evenness (Figure 4
These taxonomic distinctions are especially relevant for understanding the potential effects of irrigation on diversity. While irrigation allows farmers to expand the range of crop species that can be grown, species diversification can also lead to genetic (within species) specialization. In the case of irrigation, changes in the hydrologic conditions under which crop landraces developed can render these landraces less able to compete with newly introduced species or cultivars [96
In sum, crop diversity measures can increase and decrease simultaneously depending on several factors. These include crop taxonomy, the measurement techniques or diversity indices being used (e.g., richness vs. evenness), the spatial scales of analysis, and the levels of data aggregation (e.g., farm level, regional level, national level) [16
]. Each of these factors has implications for how the different drivers of crop diversity ultimately impact biodiversity and ecosystem services [101
]. Therefore, we caution against fully embracing irrigation expansion as a means of enhancing crop diversification without better understanding the full range of potential effects at different taxonomic, spatiotemporal, and functional levels.