Functional trait-mediated limiting similarity shapes weed biomass acquisition, fecundity, and reproductive phenology

Abstract

According to the theories of limiting similarity and life history trade-offs, (i) the probability of species coexistence decreases as interspecific similarity increases, and (ii) stronger competition intensity may accelerate reproductive phenology. Based on these theories, we predict that (i) weed biomass and fecundity will decrease and (ii) reproductive phenology will increase when competing with functionally similar crops. To test these predictions, we evaluated biomass acquisition, fecundity, time of anthesis, and seed shattering of four eudicot and two monocot weed species growing in two crops [Glycine max (L.) Merr. and Zea mays L.]. We then used a trait-based approach to assess limiting similarity and its feasibility as a predictor of crop-weed competitive outcomes. Results revealed greater functional similarity between G. max and four eudicot weeds than between Z. mays and these species. In accordance with our first prediction, weed biomass and fecundity of eudicot species were, on average, 25 and 11% lower in G. max than in Z. mays, respectively, suggesting higher competitive interactions. In contrast, the functional similarity between Z. mays and the two monocot weeds was higher than that between G. max and these two monocot weeds. Accordingly, biomass and fecundity of these species were, on average, 35 and 13% lower, respectively, in Z. mays than in G. max. In accordance with our second prediction, greater crop-weed functional similarity was positively correlated with seed shattering across all weed species. Our results demonstrate that crop-weed functional similarity may be a useful indicator for predicting weed biomass, fecundity, and seed shattering under competitive filtering (e.g., crops and cover crops), which are fundamental components of ecologically based weed management. Therefore, a trait-based approach in agroecosystems could provide mechanistic insights into crop-weed interactions, inform management priorities for problematic weed species, and help increase biodiversity and ecosystem services, supporting the design of resilient cropping systems.

Introduction

Competition is both a driver and a consequence of evolution and remains a central tenet of fundamental and applied ecology^1–3. Over the last four decades, despite extensive research assessing the effects of interspecific competition on the fitness of interacting species, the extent to which its intensity varies among species remains an enduring and partially answered question that requires further investigation. The earliest attempt to address this question dates back to Charles Darwin⁴, who observed that competitive interactions tend to be stronger among closely related species due to their similar ecological requirements. Almost a century later, this idea was formalized as the theory of limiting similarity, an extension of the competitive exclusion principle⁵. This theory posits that coexisting species must differ in their ecological niches to a degree beyond a critical level of similarity to avoid competitive exclusion, and it implies a positive correlation between interspecific similarity and competition intensity⁶. Early efforts to define species similarities relied on a subjective set of morphological characteristics that often had limited ability to accurately reflect resource-use efficiency⁷. Recent advances in trait-based analysis, along with the availability of comprehensive databases, have enabled objective and quantitative assessments of interspecific similarity^8,9. Within the limiting similarity framework, species with similar functional traits (i.e., measurable characteristics of an organism that determine its fitness and performance)^10,11, are expected to exhibit higher niche overlap, resulting in stronger competitive interactions than those with more functionally divergent traits¹².

While studies have reported evidence of limiting similarity in natural ecosystems^13,14, empirical and experimental insights into how this framework can be used to predict crop-weed competition remain sparse. In this study, we integrated two concepts associated with limiting similarity and trait-based analysis to predict crop-weed competitive outcomes. First, according to the limiting similarity theory, a higher crop-weed functional similarity may lead to increased resource overlap, resulting in reduced weed biomass acquisition and fecundity⁵. Second, functional traits important for resource acquisition, such as plant height, leaf economic spectrum, and root architecture, can be used to predict competitive outcomes between crops and weeds¹⁵.

Among reproductive traits, time of anthesis and seed shattering, which are associated with the ability of a plant to disperse its seeds, play crucial roles in the spread of wild and weedy species¹⁶. The impact of interspecific competition on plant fitness is well established¹⁷, and annual weedy species may also adjust their reproductive timing in response to competition intensity¹⁸; however, debate exists regarding the pattern of species responses. For example, plant strategy theory predicts a delayed reproductive phenology under high competition¹⁹. In contrast, the life history theory suggests accelerated reproduction in highly competitive environments²⁰. However, how weed reproductive phenology may vary across different crops remains largely underexplored. Integrating limiting similarity with trait-based approaches to evaluate weeds under different competitive filtering may help elucidate the underlying ecological processes that regulate weed reproductive phenology. From an applied viewpoint, understanding the outcomes of crop-weed competition can aid integrated weed management while increasing biodiversity, ecosystem services, and agroecosystem resilience²¹.

We conducted field experiments to evaluate the impact of two functionally different crops: (1) corn (Zea mays L., monocot) and (2) soybean [Glycine max (L.) Merr., eudicot], on biomass acquisition, fecundity, and reproductive phenology of annual eudicot and monocot weed species. We hypothesized that (1) according to the limiting similarity theory, crop-weed functional similarity is positively correlated with interspecific competition intensity, and (2) based on life-history theory, annual weed species exhibit earlier and faster reproductive phenology in response to increasing interspecific competition intensity.

Materials and methods

Study site

Field experiments were conducted during the 2022 and 2023 growing seasons (May-October) in conventionally managed Z. mays-G. max rotations at the Montana State University, Southern Agricultural Research Center, Huntley, Montana, USA. In each year, both crop phases were present in the same field, and the crop phases were arranged in separate blocks. The total experimental area, including both crop phases, was 2970 m². Both monocot and eudicot weed species were present within each crop phase. The soil at the study site was characterized as Fort Collins clay loam (fine-loamy, mixed, superactive, mesic Aridic Haplustalfs) with an organic matter content of 2.5% and a pH of 7.5–7.8, as measured by regular soil testing (Agvise Laboratory, Northwood, North Dakota, United States). The ground preparation included one pass of disk harrowing in the fall, followed by field cultivation and leveling in the spring before planting. Glyphosate- and glufosinate-resistant Z. mays (Hybrid: P9492YHR) and glyphosate-resistant G. max (Hybrid: REA-RX0721) were planted following regional agronomic recommendations on 20th May and 24th May in 2022 and 2023, respectively. Both crops were fertilized with nitrogen, phosphorus, and potash according to regional agronomic recommendations. Irrigation was supplied as needed using an overhead linear system.

Experimental setup and weed plant establishment

Weed individuals used in the field experiments were naturally emerging plants that germinated in situ from the resident seedbank within the experimental field. The study was arranged as a completely randomized design with 10 replications, with each weed plant treated as an individual replicate. Because parentage was not controlled, potential maternal and genetic effects among individuals within species should be considered a limitation of the study. Tested weed species were barnyard grass [Echinochloa crus-galli (L.) Beauv.], common lambsquarters (Chenopodium album L.), green foxtail [Setaria viridis (L.) Beauv.], kochia [Bassia scoparia (L.) A. J. Scott], Powell amaranth (Amaranthus powellii S.Wats.), and redroot pigweed (Amaranthus retroflexus L.). When weeds reached a height of 5–8 cm, 10 plants of each weed species per crop were randomly selected, marked with color-coded flags, and designated as focal plants. Random selection of focal plants was used to account for spatial bias within the field. To standardize the competition intensity, focal plants were at the center of the row spacing, which was 70 cm for Z. mays and 40 cm for G. max. All other non-crop plants surrounding the focal plants were manually removed. Afterward, glyphosate (1.3 kg ae ha^− 1) was applied using a handheld CO₂-pressurized sprayer calibrated to deliver a 187 L ha^− 1 spray volume, avoiding direct contact with the focal plants. Subsequently, focal plants were visually assessed for phytotoxicity symptoms for three weeks, which may have occurred through herbicidal drift. However, no symptoms were observed.

Trait data collection

Seven functional traits associated with resource acquisition (see Table 1 for their significance) were considered to quantify the functional similarity between crops and weeds. Two traits were measured under field conditions. The other five traits were obtained from plant databases for the tested crop and weed species⁹(Table 2). Trait descriptions are as follows:

Table 1.

Traits used to evaluate species similarity and their functional significance.

Traits	Functions	Functional significance
Plant height	Light-capturing ability	Taller plants intercept more light. Improved photosynthesis in a competitive environment.
Specific leaf area		A higher specific leaf area enables greater light absorption per unit of leaf biomass. Reflects the ability of plants to optimize light use in resource-limited environments.
Leaf dry matter content	Resource use efficiency	Greater leaf dry matter content indicates a higher proportion of structural material. Implies a more efficient use of available resources for growth and defense.
Leaf nitrogen content	Nitrogen use efficiency	High leaf nitrogen content positively correlates with increased growth and reproduction rates. Plants with high leaf nitrogen can optimize nitrogen resource use for faster growth and enhanced competitiveness.
Photosynthesis rate	Energy acquisition	Reflects the plant’s efficiency in converting light energy into chemical energy. Higher photosynthetic rates enable plants to accumulate more energy for growth and reproduction, increasing their competitiveness.
Rooting depth	Soil resource acquisition	Reflects the ability of plants to access water and nutrients from soil profiles. Plants with differential root depths may partition belowground resources more efficiently.
Rooting system		The root architecture impacts resource acquisition efficiency. Fibrous root systems typically allow efficient nutrient and water uptake from the upper soil profiles. Taproot systems facilitate access to deep soil layers, capturing resources unavailable to shallow-rooted plants.

Table 2.

Species × trait matrix used to assess crop-weed functional similarity using Gower’s distance.

Functional traitsa	Unitb	Speciesc
		Amaranthus powellii	Amaranthus retroflexus	Bassia scoparia	Chenopodium album	Echinochloa crus-galli	Glycine max	Setaria viridis	Zea mays
Plant height*	m	1.0	1.0	1.5	1.4	0.7	0.9	0.8	2.0
Specific leaf area	mm2 mg− 1	16.57	16.57	21.50	36.70	27.52	25.51	26.89	26.60
Leaf dry matter content*	mg g− 1	246.51	245.86	260.81	239.57	229.79	267.5	221.82	224.28
Leaf nitrogen content	g m− 2	1.5	1.5	1.0	1.9	0.6	1.7	0.4	1.3
Photosynthesis rate	µmol m− 2 s− 1	20.1	20.1	22.7	22.3	15.1	23	19.3	18.4
Rooting depth	m	2.4	2.4	2.1	2.2	1.0	2.4	0.5	1.2
Rooting systemd	-	TR	TR	TR	TR	FR	TR	FR	FR

^aFunctional traits with an asterisk (*) were measured in field conditions, and the others were derived from the TRY plant global database⁹ and local weed identification manuals.

^bAll units are based on Cornelissen, et al.²².

^cAmaranthus retroflexus was used as a model species for Amaranthus powellii.

^dAbbreviations: TR, taproots; FR, fibrous roots.

• i.Plant height: At 28 days after crop physiological maturity, the height of each focal plant was measured from the soil surface using a cm scale. For G. max and Z. mays, ten plants were randomly selected from the field, and their height was measured using a similar approach. According to Cornelissen et al.²², inflorescences or panicles were excluded from the measurements if projected above the foliage.
• ii.Leaf dry matter content: For weed species, ten fully expanded healthy leaves were collected from each adult focal plant. For crop species, ten fully developed and healthy leaves per plant were collected from ten randomly selected plants in the field. The collected leaves were blotted dry with tissue paper before fresh biomass was recorded. These leaves were dried in a hot-air oven at 60 °C for 72 h and then weighed to determine dry biomass. The dry biomass (mg) and fresh biomass (g) ratios were expressed as mg g^− 1.
• iii.Specific leaf area, leaf nitrogen content, and photosynthesis rate: These traits were derived from the TRY plant global database⁹ for the tested crop and weed species. Data for A. powellii were unavailable; therefore, A. retroflexus was used as a model species because of its similarity in phylogenetic and morphological characteristics²³.
• iv.Maximum rooting depth and rooting systems: Information on rooting depth (m) and qualitative description of the rooting system (taproot vs. fibrous) was obtained from regional weed and crop identification manuals and the United States Department of Agriculture databases²⁴. Similar to other derived traits, A. retroflexus was used as a model species for A. powellii because of the unavailability of data.

In-field weed demographics data collection

The time of anthesis for each weed species was recorded for all plants, and plastic structures (1 m²) were established around them to capture shattered seeds using thermoformed trays (1020 Heavy Duty, CN-FNHD-X1, Greenhouse Megastore, Danville, IL, USA). From the first seed rain onwards, shattered weed seeds were collected using a handheld battery-operated vacuum cleaner at weekly intervals up to 28 days after crop physiological maturity. At each collection time, the number of shattered seeds was estimated by multiplying the weight of shattered seeds by 100 and subsequently dividing it by the weight of 100 seeds for each weed species. These data were converted to cumulative seed shattering (%). At 28 days after crop physiological maturity, the retained weed seeds were manually harvested, and their numbers were estimated using the approach described above. Focal plants were then cut from the soil surface using a garden pruner, placed in individual paper bags, and dried in a hot air oven at 60 °C for 72 h to determine their dry biomass. Fecundity per plant was calculated by summing the shattered and retained seeds.

Statistical analysis

R Studio (v 4.2.1, R Development Core Team) was used for analysis and visualization. To test our first hypothesis, pairwise interspecific functional similarity was estimated by Gower’s distance, calculated based on the species × trait matrix (Table 2) using the ‘gowdis’ function of the “FD” package^25,26. Gower’s distance ranged from 0 to 1, representing the highest and lowest crop-weed functional similarities, respectively. The Gower’s distances between crops and weeds were visualized using a bipartite network graph.

Dry biomass (g plant^− 1) and fecundity (seeds plant^− 1) data were fitted into two separate generalized linear mixed models using the “glmmTMB” package with a default Gaussian distribution. Initially, full models were developed using crop type, weed species, years, and replications as explanatory variables. No differences were observed between the two years (p > 0.05); therefore, the models were reduced by removing nonsignificant variables using the parsimony principle based on the Akaike Information Criterion. Crop type and weed species were included as fixed effects, and replication was included as a random effect to capture the inherent variability across experimental replicates. The final corresponding models were (i) dry biomass ~ crop types + weed species + crop types*weed species + (1|replication) and (ii) fecundity ~ crop types + weed species + crop types*weed species + (1|replication). Data diagnostics included simulating residuals and assessing multicollinearity using the “DHARMa” and “performance” packages, respectively. Means were separated by Tukey’s honestly significant difference test at α = 0.05 and α = 0.1 using the “multcomp” and “emmeans” packages.

To assess trait-mediated differences in crop-weed competition, a weed suppression index (WSI) (Eqs. 1 and 2) and fecundity reduction index (FRI) (Eqs. 3 and 4) were estimated using pairwise comparisons of individual weed plants grown in the two crop environments. For each weed species, 20 individual plants were evaluated in each crop across the two study years; therefore, the 400 values correspond to all possible pairwise combinations between plants from the two crops (20 × 20 = 400) represent an exhaustive rather than random comparison. Because the same individual plants contributed to multiple pairwise contrasts, these values were treated as derived comparative observations describing relative crop effects on weed biomass and fecundity, rather than as 400 independent biological replicates. For each focal crop, higher WSI and FRI values indicate greater weed biomass suppression and fecundity reduction by designated crops, respectively. We then plotted WSI and FRI against the pairwise Gower’s distance between each weed species and their respective crop to evaluate trait-mediated crop-weed interactions. Multiple candidate regression models (e.g., linear, quadratic, and asymptotic) were fitted to examine the relationships between the indices and Gower’s distance. Final models were then selected primarily based on ecological interpretability, Akaike Information Criterion, residual diagnostics, goodness-of-fit, and R² values. The 2nd -degree polynomial regressions and simple linear regressions were retained for Z. mays and G. max, respectively, to describe the data pattern. The fitted models were used as an empirical tool and therefore do not necessarily imply the presence of underlying quadratic ecological mechanisms.

1

where, WSI_{i, j → corn} = Weed Suppression Index for Z. mays, b_{soy, i} = Biomass of the i-th weed plant in G. max; where i = 1, 2,…, 20, and b_{corn, j} = Biomass of the j-th weed plant in Z. mays; where j = 1, 2,…, 20.

2

where, WSI_{i, j → soybean} = Weed Suppression Index for G. max, b_{corn, i} = Biomass of the i-th weed plant in Z. mays; where i = 1, 2,…, 20, and b_{soy, j} = Biomass of the j-th weed plant in G. max; where j = 1, 2,…, 20.

3

where, FRI_{i, j → corn} = Fecundity Reduction Index for Z. mays, f_{soy, i} = Fecundity of the i-th weed plant in G. max; where i = 1, 2,…, 20, and f_{corn, j} = Fecundity of the j-th weed plant in Z. mays; where j = 1, 2,…, 20.

4

where, FRI_{i, j → soybean} = Fecundity Reduction Index for G. max, f_{corn, i} = Fecundity of the i-th weed plant in Z. mays; where i = 1, 2,…, 20, and f_{soy, j} = Fecundity of the j-th weed plant in G. max; where j = 1, 2,…, 20.

A complementary principal component analysis (PCA) was performed on scaled numerical traits to explore multivariate trait overlap using the ‘PCA’ function of the ‘FactoMineR’ package²⁷. A categorical trait (e.g., rooting system) was omitted from the PCA. In trait space, species coordinates were used to construct polygons describing functional boundaries of eudicots and monocots using the “concaveman” package. Additionally, the vectors of relative trait contribution (% to PC1 + PC2) were overlaid on the ordination spread to explore the influence of individual traits on species separation, which could potentially lead to the trait hierarchy (i.e., certain dominant traits disproportionately drive crop-weed competition).

To test our second hypothesis, cumulative growing degree days (GDD) (°C) were calculated for both crops (Eq. 5) from the planting date to the harvest date. Species-wise cumulative seed shattering (%) and GDD were used to calculate the GDD required for 10% seed shattering (S10) using the “SeedCalC” package. In addition, the seed shattering rate index (SSRi) was calculated using Eq. 6, which was modified from the germination rate index described by Silva, et al.²⁸. The time of anthesis, S10, and SSRi data were fitted to separate generalized linear mixed models using the procedure described above. Data diagnostics and mean separation were conducted as described above.

5

Where, i = day of interest, j = index of summation; from day 1 to day I, T_{max, j} = maximum temperature on day j, T_{min, j} = minimum temperature on day j, and T_base = base temperature (10 °C)

6

.

S_i = Cumulative seed shattering percentage at each recorded GDD interval.

ΔGDD_i = Difference between consecutive GDD values.

∑ΔGDD_i= Total accumulated GDD over the observation period.

Results and discussion

Impact of functional similarity on crop-weed competition

Functional similarity patterns varied between the two crops and the six weed species studied. Between G. max and the four studied eudicot weed species (A. powellii, A. retroflexus, B. scoparia, and C. album), Gower’s distance values were relatively low, ranging from 0.24 to 0.27, indicating greater functional similarity (Fig. 1). In comparison, Gower’s distances between G. max and two studied monocot species (E. crus-galli and S. viridis) were relatively high (0.59 and 0.58, respectively), suggesting lower functional similarity (Fig. 1). Zea mays, conversely, exhibited lower functional similarity with A. powellii, A. retroflexus, B. scoparia, and C. album, with Gower’s distance ranging from 0.54 to 0.59, than with E. crus-galli and S. viridis, with Gower’s distances of 0.33 and 0.28, respectively (Fig. 1).

Fig. 1.

Bipartite network graph representing the pairwise Gower’s distance between crops (center) and weed species (periphery), in which connecting lines with different colors and thicknesses represent functional similarities based on Gower’s distance (i.e., lower values suggest higher functional similarity.

Based on Tukey’s HSD at α = 0.05 (unless specified otherwise), biomass acquisition of A. powellii, A. retroflexus, B. scoparia, and C. album was 23, 26, 27, and 24% lower, respectively, in G. max than in Z. mays (Fig. 2A). In contrast, E. crus-galli and S. viridis produced 37 and 34% lower biomass respectively, in Z. mays than in G. max (Fig. 2A). Similarly, A. powellii, A. retroflexus, B. scoparia, and C. album showed 11, 24, 4 (α = 0.1), and 6% (α = 0.1) lower fecundity, respectively, in G. max than in Z. mays (Fig. 2B). In comparison, the fecundity of E. crus-galli and S. viridis was 6 and 21% lower, respectively, in Z. mays than in G. max (Fig. 2B). These results suggest that differences in functional similarity consistently corresponded to biomass acquisition and fecundity patterns in the studied weed species across both crops.

Fig. 2.

(A) Biomass (g plant^− 1) and (B) fecundity (seeds plant^− 1) of weed species grown in competition with either Glycine max or Zea mays. Tukey’s honestly significant differences are indicated by * at α = 0.05 and ** at α = 0.1.

According to the limiting similarity theory, lower Gower’s distance (i.e., greater trait similarity) is associated with higher competition intensity²⁷. As such, WSI and FRI values should decrease as Gower’s distance increases. Consistent with this expectation, we observed negative relationships across all tested scenarios (Fig. 3). In Z. mays, WSI showed a strong negative relationship (linear coefficient b = -20.451x) initially and then flattened at higher Gower’s distance (quadratic coefficient a = 18.629). In contrast, FRI in Z. mays consistently exhibited a negative relationship (y = 1.006 – 2.021x − 3.928 x ²) with Gower’s distance between and within weed functional groups. In G. max, WSI showed a strong linear negative relationship (y = 1.915 – 2.127x) with Gower’s distance between and within weed functional groups. However, for FRI ~ Gower’s distance in G. max, negative relationships were only apparent between weed functional groups (y = 1.425 – 1.037x), but such trends were absent within functional groups. While the relatively low number of tested species precludes us from testing trends within weed functional groups, negative relationships were apparent in five out of eight scenarios: monocots in Z. mays for WSI, monocots and eudicots in Z. mays for FRI, and monocots and eudicots in G. max for WSI (Fig. 3).

Fig. 3.

Weed suppression and fecundity reduction indices were negatively correlated with Gower’s distance. A 2nd -degree polynomial regression was used to describe the relationship between the weed suppression index/fecundity reduction index and Gower’s distance in Zea mays. Such relationships were described by the simple linear regression in G. max. The R² values represent the coefficient of determination.

Overall, in partial agreement with our first hypothesis (i.e., limiting similarity)^5,7, positive relationships between competitive outcomes (weed suppression and fecundity reduction) and functional similarity were consistent across crop-weed functional groups but may not generalize within functional groups. In agroecosystems, this suggests that limiting similarity may shape crop-weed competitive interactions at broader functional contrasts. However, additional mechanisms could be a driving factor within narrow trait space, such as trait hierarchy (discussed below). A handful of previous studies have formally evaluated the existence of a link between interspecific competition and functional traits^12,15. To the best of our knowledge, our results represent the first attempt to expand this framework to predict crop-weed interactions. From an applied perspective, such insights would be useful for designing trait-informed integrated weed management programs that optimize functional diversity in crop rotations²⁹.

A complementary PCA-based trait overlap analysis provided additional insights into the relative contributions of individual traits to species separation (Fig. 4). Crop and weed coordinates were distributed along the PC1 and PC2 axes, reflecting functional variation among the tested species. As expected, G. max clustered closely with broadleaf weed species within the functional boundaries of eudicots. Although Z. mays occupied a distinct region in a trait-space, likely due to two divergent traits (discussed below), it aligned with the two tested grass weeds and formed a separate monocot polygon, with no overlap into the functional boundaries of eudicots. Overall, traits such as specific leaf area and plant height had the highest contribution (> 35%) to species separation, followed by rooting depth, leaf dry matter content, leaf nitrogen content, and photosynthesis rate (Fig. 4). Echinochloa crus-galli and S. viridis were primarily separated from G. max and other weed species by leaf dry matter content, leaf nitrogen content, photosynthesis rate, and rooting depth along the PC1 (Fig. 4), suggesting that these traits were more influential in the separation of these weeds from G. max and other eudicot weed species. The separation of Z. mays from E. crus-galli and S. viridis along PC2 is due to their plant height and specific leaf area (Fig. 4). Similarly, C. album was separated from the other species along PC2, highlighting the role of specific leaf area in defining its divergence from G. max and other eudicot weed species.

Fig. 4.

Principal component analysis-based trait overlap, in which colored polygons and arrows delineate functional boundaries of different groups (e.g., crop, weeds, eudicots, and monocots) and relative contribution of individual traits to species separation, respectively. Trait abbreviations: LDMC; leaf dry matter content, LNC; leaf nitrogen content, PH; plant height, PSR; photosynthesis rate, RD; rooting depth, and SLA; specific leaf area.

In summary, our PCA-based trait overlap approach suggests that, among species that appear functionally similar based on pairwise metrics, underlying divergence in specific traits may exist, potentially contributing to trait hierarchies in which these traits disproportionately drive competitive outcomes^12,30,31. In agreement with our results, such trait hierarchies were reported in a mesocosm experiment in which interspecific competition among eight annual plant species was predominantly driven by plant height and specific leaf area³⁰. In a grassland ecosystem, specific root length, root-to-shoot ratio, and leaf nitrogen content were the most prominent drivers of competitive interactions among 13 tested traits¹². A direct quantification of trait hierarchies under field conditions was beyond the scope of this study; however, our PCA-based analysis suggests that such assessments could provide mechanistic insights into the management of agricultural weeds. While our results suggest that integrating crop-weed trait similarity and trait hierarchy may serve as a predictive and mechanistic framework^32,33, we acknowledge that caution is warranted when extrapolating these findings to other agroecosystems, given that a limited set of weed species across two crops was tested. In this context, further investigation into trait-mediated crop-weed interactions, with a broader range of species and functional traits across different agroecosystems, would help validate the generality of our results.

Impact of crop-weed functional similarity on weed reproductive phenology

In relation to our second hypothesis, crop-weed functional similarity did not affect weed anthesis but influenced weed seed shattering phenology. S10 was, on average, 54 °C GDD lower for A. powellii, A. retroflexus, B. scoparia, and C. album under the functionally similar G. max compared to Z. mays (Fig. 5). However, this trend was not observed for E. crus-galli and S. viridis.

Fig. 5.

Reproductive phenology of six weed species growing in two crops (Zea mays and Glycine max). Response traits included time of anthesis (°C growing degree days to the beginning of anthesis), S10 (°C growing degree days required for 10% cumulative seed shattering), and seed shattering rate index (SSRi, see main text, unitless). For SSRi, a bar plot is used because of the narrow standard errors of the mean, represented by black error bars. Asterisks (*) represent significant differences in traits between crop identities based on Tukey’s honestly significant difference test at α = 0.05.

SSRi values capture the rate of seed shattering phenology, with greater values indicating faster seed shattering. In agreement with our second hypothesis, SSRi values were consistently greater across all tested weed species growing in competition with functionally similar crops than the values observed in dissimilar crops. For example, SSRi for A. powellii, A. retroflexus, B. scoparia, and C. album ranged from 5 to 25 in functionally similar G. max but ranged from 2 to 20 in functionally dissimilar Z. mays (Fig. 5). Similarly, E. crus-galli and S. viridis showed 62 and 10 SSRi in functionally similar Z. mays, respectively, which were higher than 32 and 8 SSRi observed in the functionally dissimilar G. max (Fig. 5). These findings suggest that higher crop-weed functional similarity was consistently associated with faster seed shattering across the species tested. Although direct empirical evidence for accelerated seed shattering resulting from increased competition intensity is limited, a previous study demonstrated a negative correlation between seed shattering and individual plant fitness across seven annual weed species³⁴. In that study, it is plausible that interspecific competition may have affected the fitness (i.e., biomass acquisition) and seed shattering of individual weed plants. In our study, a possible explanation is that weed species experiencing stronger competitive filtering may exhibit earlier and/or faster reproductive transitions, a pattern referred to as a competitive annual strategy³⁵.

According to the theory of life-history trade-offs^19,36,37, species experiencing intense competition invest disproportionately more in reproduction at the expense of vegetative growth. In particular, Bonser²⁰ observed that annual plant species show higher reproductive efficiency in response to enhanced competition intensity. Our finding of faster seed shattering in weed species under functionally similar crops expands on Bonser²⁰ and supports the predictions of life-history theory, partially aligning with our second hypothesis. Consistent with a competitive annual strategy, we observed lower weed biomass and higher SSRi under functionally similar crops, potentially increasing the prospects for progeny establishment in subsequent seasons^19,35,38. However, we acknowledge that the underlying mechanisms remain uncertain, and physiologically driven research on resource allocation strategies in weed species under various competitive filters could provide insights into the observed pattern.

Conclusion

Our study provides evidence that limiting similarity at both species and trait levels may be used to infer weed biomass acquisition, fecundity, and reproductive phenology in agroecosystems. Results indicate that functional similarity between crops and weeds is positively related to competition intensity and may be a useful indicator in predicting the intensity of crop-weed competition. Untangling the roles of trait similarity and/or hierarchy as drivers of competition among multiple crops and weed species across different agroecosystems, and the mechanisms underlying these patterns, would further refine the predictive capabilities of a trait-based analysis. We obtained several functional characteristics from trait databases, an approach that does not account for weed phenotypic plasticity in specific niches^39,40. We also demonstrated that crop-weed functional similarity was associated with weed reproductive phenology, particularly seed shattering, consistent with the broader theory of life-history trade-offs²⁰. While functional characteristics quantified from field measurements may enhance the predictive power of the current approach, trait-based analyses with data derived from databases have been successfully used to study weed community assembly in agroecosystems^39,41. Our results emphasize that integrating limiting similarity with a trait-based approach has far-reaching ecological implications and may serve as a predictive framework for developing integrated management strategies to enhance ecosystem services from weed floras⁴². From an applied standpoint, decisions on cover crop selection, cultivar choice, and crop rotations could be informed by functional characteristics specific to target weed communities across diverse agroecosystems.

Author contributions

Conceptualization: H.S.D., F.D.M., and L.S.S.; Preparing original draft: H.S.D.; Data collection: H.S.D.; Data analysis: H.S.D. and F.D.M.; Reviewing: F.D.M. and L.S.S.; Funding acquisition: L.S.S.

Funding

The authors acknowledge the research funding from the USDA-NIFA Hatch Project (MONB00932) and the Montana Wheat and Barley Committee.

Data availability

Raw data from this study will be made available upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.