Introduction
Functional trait approaches have advanced our understanding of biodiversity responses to environmental change in tropical and subtropical ecosystems. However, these studies predominantly aggregate individuals into species-level means, thereby obscuring intraspecific variation that often contains critical ecological signals (Violle et al. Reference Violle, Enquist, McGill, Jiang, Albert, Hulshof, Jung and Messier2012). Within species, males and females frequently exhibit divergent morphologies, physiological traits, and ecological tolerances, yet sex-specific trait variation remains underrepresented in tropical ecology research (Blanckenhorn Reference Blanckenhorn2005; Stillwell et al. Reference Stillwell, Blanckenhorn, Teder, Davidowitz and Fox2010). Ignoring these differences may mask sex-specific habitat requirements and population-level limitations, particularly in degraded forests where restoration success depends on understanding demographic vulnerabilities (Hangartner et al. Reference Hangartner, Sgrò, Connallon and Booksmythe2022).
Dung beetles provide an ideal system for examining sex-dependent functional responses in tropical and subtropical ecosystems (deCastro-Arrazola et al. Reference deCastro-Arrazola, Andrew, Berg, Curtsdotter, Lumaret, Menéndez, Moretti, Nervo, Nichols, Sánchez-Piñero, Santos, Sheldon, Slade and Hortal2023; de Oliveira Ribeiro et al. Reference de Oliveira Ribeiro, Camargo and Frizzas2024). Males typically exhibit enlarged pronota and protibiae that are adapted for combat and mate-guarding, whereas females possess broader abdomens along with specialised digging structures for brood provisioning (de Oliveira Ribeiro et al. Reference de Oliveira Ribeiro, Camargo and Frizzas2024; McCullough and Simmons Reference McCullough and Simmons2016). These divergent selective regimes – specifically, sexual selection favouring male competitive ability versus natural selection optimising female reproductive efficiency – are expected to generate contrasting patterns of phenotypic integration (Armbruster et al. Reference Armbruster, Pélabon, Bolstad and Hansen2014; Klingenberg Reference Klingenberg2008). According to theoretical predictions, sexual selection may favour modularity (the compartmentalisation of combat-related traits) in males, thereby allowing functional specialisation without compromising other performance axes, whereas natural selection may favour integration (i.e. tight trait covariation) in females to coordinate digging, provisioning, and related functions (Friedman et al. Reference Friedman, Remeš and Economo2019; Klingenberg Reference Klingenberg2014). As keystone ecosystem engineers in tropical and subtropical forests, dung beetles mediate nutrient cycling and secondary seed dispersal, making the recovery of their functional roles a priority in degraded landscapes (deCastro-Arrazola et al. Reference deCastro-Arrazola, Andrew, Berg, Curtsdotter, Lumaret, Menéndez, Moretti, Nervo, Nichols, Sánchez-Piñero, Santos, Sheldon, Slade and Hortal2023; Nichols et al. Reference Nichols, Spector, Louzada, Larsen, Amezquita and Favila2008).
Although single-trait analyses effectively identify dimorphic traits, they cannot capture how multiple traits covary as integrated phenotypes. Phenotypic integration theory posits that adaptation involves coordinated variation among functionally related traits, which are organised either as tight modules reflecting functional specialisation or as integrated networks reflecting overall trait coordination (Armbruster et al. Reference Armbruster, Pélabon, Bolstad and Hansen2014; Friedman et al. Reference Friedman, Remeš and Economo2019; Hangartner et al. Reference Hangartner, Sgrò, Connallon and Booksmythe2022). Trait network analysis quantifies these coordination patterns through topological parameters – network density (overall integration), modularity (the degree of functional compartmentalisation), and average path length (coordination efficiency among traits) – providing a framework for comparing sex-specific functional architectures (He et al. Reference He, Li, Liu, Xu, Li, Zhang, He, Tang, Han, Ye, Xiao, Yu, Liu, Sun, Niu, Li, Sack and Yu2020; Li et al. Reference Li, Liu, Sack, Xu, Li, Zhang and He2022). Although these approaches have advanced plant functional ecology, their application to animal sexual dimorphism, particularly in tropical and subtropical invertebrates, remains limited.
Environmental filtering theory further suggests that males and females may respond differently to habitat-related environmental gradients including vegetation structure, litter characteristics, soil moisture, and topography, owing to sex-specific resource requirements and movement patterns (de Oliveira Ribeiro et al. Reference de Oliveira Ribeiro, Camargo and Frizzas2024; Hangartner et al. Reference Hangartner, Sgrò, Connallon and Booksmythe2022; Salomão et al. Reference Salomão, Cerqueira, Gomes, González-Tokman, Maia and Iannuzzi2022). Females, constrained by parental investment in brood chambers, are expected to respond to fine-scale microhabitat conditions, whereas males, engaged in mate-searching across broader territories, may respond to landscape-scale gradients (Estes et al. Reference Estes, Hearn, Snell-Rood, Feindler, Feeser, Abebe, Dunning Hotopp and Moczek2013; Manning and Ford Reference Manning and Ford2016). However, whether these sex-specific environmental associations manifest at the level of single traits, network topology, or both, remains unexplored in dung beetles.
To address this central question, we analysed sex-dependent morphological variation in the dung beetle Onthophagus atripennis (Coleoptera: Scarabaeidae: Scarabaeinae) using both single-trait and multi-trait network approaches. We tested three hypotheses: (1) males and females exhibit significant sexual dimorphism in morphological traits, with males being larger than females in both body size and functionally important structures; (2) trait network structure differs between sexes, with females showing higher network connectivity reflecting integration strategies and males showing higher modularity reflecting functional specialisation; and (3) males and females respond to distinct environmental drivers, leading to sex-specific trait–environment and network–environment relationships. By integrating single-trait analyses with trait network approaches, we sought to determine whether sex differences in O. atripennis manifest not only in mean trait values but also in the broader organisation of phenotypic integration and environmental responses. Collectively, this study provides a baseline for evaluating ecological restoration and soil fauna functional recovery in tropical and subtropical forests.
Materials and methods
Study area
Fieldwork was conducted from August to December 2021, a limited period owing to COVID-19 restrictions, at the Bore Temple State-owned Experimental Forest Farm and surrounding areas in Dujiangyan, Sichuan Province, southwestern China (31°04′–31°05′ N, 103°42′–103°43′ E; 600–1,000 m a.s.l.). The region experiences a subtropical humid monsoon climate characterised by frequent fog, a mean annual temperature of 15°C, and annual precipitation of 1,200–1,800 mm (Chen Reference Chen2000). The landscape comprises secondary forest dominated by evergreen and deciduous broad-leaved species, which reflect severe historical fragmentation and disturbance caused by the 2008 Wenchuan earthquake (Xiao et al. Reference Xiao, Wang, Zhang and Ma2002; Yang et al. Reference Yang, Xie, Tao, Li and Xiao2015). We established 14 plots across a range of heterogeneous forest conditions (canopy cover 40–95%, litter depth 5–8 cm; Figure 1; Table S1).
Location of the study area and sampling sites in Dujiangyan, Sichuan Province, southwestern China. The inset map indicates the position of Dujiangyan within Sichuan Province. Sampling plots are shown as black dots within the Dujiangyan boundary (dashed line). Elevation ranges from 590 to 4239 m.
Sampling methods
Dung beetles were collected using baited pitfall traps. At each plot, we established one 20 m × 40 m quadrat, which was divided into eight 10 m × 10 m sub-quadrats (Figure S1). Plastic cup traps (25 ml) baited with 20 g of fresh pig dung, human faeces, or decayed fish fermented with cheese were placed at sub-quadrat vertices (Larsen and Forsyth Reference Larsen and Forsyth2005; Ong et al. Reference Ong, Slade and Lim2020). Traps contained saturated NaCl solution (one-third volume) to preserve specimens and were equipped with drainage holes and plastic covers to prevent contamination. Traps remained in the field for 48 hours, after which specimens were preserved in 75% ethanol.
Specimens were identified morphologically and verified by taxonomists. Onthophagus atripennis was confirmed as the dominant dung beetle species in the study area, accounting for 299 individuals (150 females and 149 males) across the 14 sampling plots (Table S2). Trait network analysis requires stable correlation estimates, which necessitates at least five individuals per sex per plot to ensure reliable network topology (He et al. Reference He, Li, Liu, Xu, Li, Zhang, He, Tang, Han, Ye, Xiao, Yu, Liu, Sun, Niu, Li, Sack and Yu2020). Consequently, network analyses were strictly restricted to seven plots, whereas the remaining seven plots with insufficient sample sizes (fewer than five individuals per sex) were excluded from network construction.
Trait measurement
Thirteen morphological traits were measured using a stereomicroscope (ST60-24B1) and a digital vernier calliper (AS0020301), and four additional proportional traits calculated (Tables 1 and 2, Figure S2). Body size (BS) was calculated as the sum of pronotum length (PronL) and elytron length (ElyL) (Radtke and Williamson 2005). Proportional functional traits were calculated from raw measurements without standardisation to retain absolute size information (Pessôa et al. Reference Pessôa, Izzo and Vaz-de-Mello2017). Functional interpretations of the morphological traits discussed in this study were inferred from previous literature on dung beetle morphology and behaviour but were not directly tested experimentally in the present study.
Morphological traits directly measured in Onthophagus atripennis
Table 1. Long description
A table with five columns and thirteen rows detailing morphological traits of Onthophagus atripennis. The columns are labeled Trait, Code, Definition, and Functional association. Row 1: Trait: Body length, Code: BL, Definition: Distance from anterior margin of clypeus to posterior margin of elytra, excluding pygidium, Functional association: Overall body size. Row 2: Trait: Head length, Code: HeadL, Definition: Distance from anterior margin of clypeus to posterior margin of head capsule, Functional association: Sensory and feeding apparatus. Row 3: Trait: Head width, Code: HeadW, Definition: Maximum dorsal width of head capsule, Functional association: Muscle attachment for mandibular movement. Row 4: Trait: Pronotum length, Code: PronL, Definition: Median dorsal length of pronotum, Functional association: Female investment in digging. Row 5: Trait: Pronotum width, Code: PronW, Definition: Maximum dorsal width of pronotum, Functional association: Male combat structures. Row 6: Trait: Elytron length, Code: ElyL, Definition: Median length of elytra, Functional association: Flight capability. Row 7: Trait: Elytron width, Code: ElyW, Definition: Maximum dorsal width of abdomen (elytra), Functional association: Body protection. Row 8: Trait: Protibia length, Code: ProTiL, Definition: Length of fore tibia, Functional association: Female digging and brood care. Row 9: Trait: Protibia width, Code: ProTiW, Definition: Maximum width of fore tibia, Functional association: Soil manipulation force. Row 10: Trait: Mesotibia length, Code: MesoTiL, Definition: Length of mid tibia, Functional association: Walking and balance. Row 11: Trait: Mesotibia width, Code: MesoTiW, Definition: Maximum width of mid tibia, Functional association: Support stability. Row 12: Trait: Metatibia length, Code: MetaTiL, Definition: Length of hind tibia, Functional association: Propulsion and digging. Row 13: Trait: Metatibia width, Code: MetaTiW, Definition: Maximum width of hind tibia, Functional association: Soil compaction.
Functional associations are inferred from previous studies on dung beetle morphology and functional traits and were not directly tested in the present study.
Functional trait ratios used to assess sex-specific morphological responses to forest degradation in Onthophagus atripennis
Table 2. Long description
A table with four rows and four columns comparing morphological trait ratios. Column headers are Ratio index, Formula, Functional interpretation, and Sex-specific hypothesis. Row 1: Pronotum length-to-width ratio (PronL/W), PronL/PronW, Higher values have been interpreted as weaker flight capability but stronger digging/burial ability (Pessoa et al. 2017; Perrin et al. 2021), Females > males (digging investment). Row 2: Relative pronotum size (PronL/BS), PronL/(PronL+ElyL), High values have been interpreted as greater investment in dispersal than reproduction; low values indicate greater investment in reproduction than dispersal (Karlsson and Wickman 1989; Attisano and Kilner 2015), Females > males (trade-off strategy). Row 3: Elytron length-to-width ratio (ElyL/W), ElyL/ElyW, Higher values have been interpreted as slender body form and enhanced tolerance to harsh environments (Linz et al. 2016); adaptation to thick litter layers (Bui et al. 2020), Females > males (microhabitat adaptation). Row 4: Relative protibia length (ProtIiL/BL), ProTiL/BL, Higher values have been interpreted as stronger digging and dung manipulation capabilities (Macagno et al. 2016; Giménez Gómez et al. 2022), Females > males (parental investment).
1 PronL, pronotum length; PronW, pronotum width; ElyL, elytron length; ElyW, elytron width; BS, body size (calculated as PronL + ElyL); ProTiL, protibia length; BL, body length. Functional interpretations and sex-specific hypotheses are inferred from previous studies on dung beetle morphology and insect functional traits, rather than being directly tested in the present study.
Trait network construction
To explore covariation and integration patterns among morphological traits, we performed trait network analysis following He et al. (Reference He, Li, Liu, Xu, Li, Zhang, He, Tang, Han, Ye, Xiao, Yu, Liu, Sun, Niu, Li, Sack and Yu2020). Based on data from the seven valid plots, we constructed separate trait networks for females and males, yielding 14 networks (seven plots × two sexes). Nodes consisted of 18 morphological indices (13 original measurements, BS, and four proportional indices). Edges represented Pearson’s correlation coefficients between traits and were retained only when |r| ≥ 0.7 and p < 0.05, resulting in undirected binary networks (Li et al. Reference Li, Liu, Sack, Xu, Li, Zhang and He2022). Network visualisation and parameter calculations were performed in R 4.2.0 using the igraph package. Based on previous trait network studies, network density (GD) characterises the overall level of network connectivity and serves as an indicator of morphological integration. Modularity (Q) reflects the extent to which a network is divided into relatively independent modules, thereby indicating the overall degree of modular organisation. Average path length (AL) reflects network compactness and the potential coordination efficiency among traits; shorter path lengths generally indicate tighter associations and stronger coordination among traits (De Mendoza et al. Reference De Mendoza, Carril and Tambussi2025; He et al. Reference He, Li, Liu, Xu, Li, Zhang, He, Tang, Han, Ye, Xiao, Yu, Liu, Sun, Niu, Li, Sack and Yu2020). Accordingly, we compared the organisational characteristics of male and female multi-trait covariation structures with respect to overall connectivity, modularity, and network compactness, and we further examined how these network features associated with environmental factors. The formulas and ecological meanings of the trait network structural parameters are provided in Table S3.
Environmental variables
Environmental data were obtained from field measurements and remote-sensing analysis. Topographic variables included longitude, latitude, and altitude (Alt) recorded by GPS, as well as slope (Slo) measured with a handheld compass. Soil characteristics, including soil moisture (SM) and pH, were measured using a portable moisture and pH meter.
Vegetation surveys used nested quadrats. Within each plot, three 10 m × 10 m tree quadrats were randomly established to record tree species richness (TS), tree abundance (TA), diameter at breast height (DBH recorded as TD), and tree height (TH). Within each tree quadrat, two 5 m × 5 m shrub quadrats were established to record shrub species richness (SS) and shrub abundance (SA), and four 1 m × 1 m herb quadrats to record herbaceous species richness (HS), herb abundance (HA), litter thickness (LT), and litter coverage (LC). To avoid multicollinearity, we screened variables using Pearson’s correlations (threshold: |r| ≥ 0.8) and variance inflation factors (VIFs). Specifically, we excluded tree species richness (TS, r = 0.85 with TA) and litter thickness (LT, r = 0.83 with LC) as ecologically less significant within their respective pairs. The final nine retained variables (Alt, HS, HA, Slo, SA, TA, TH, LC, and SM) had VIF < 5.0.
Statistical analysis
Due to sample-size constraints, analyses were restricted to the seven plots with sufficient individuals for network construction. We performed one-way analysis of variance (ANOVA) using sex as the sole fixed factor to test for differences in morphological traits and network parameters between males and females. We applied the Benjamini–Hochberg procedure to control the Type I error rate across multiple comparisons. We calculated Pearson’s correlation coefficients, stratifying analyses by sex, to quantify associations between morphological traits and environmental variables, as well as between network parameters and environmental variables. For trait–environment analyses, we calculated sex-specific, plot-level mean values for each morphological trait to avoid pseudoreplication arising from multiple individuals sharing the same environmental context within a plot. Therefore, the sampling plot, rather than the individual, was treated as the unit of replication in all environment association analyses. This allowed us to identify sex-specific environmental drivers of trait and network variation, with statistical significance at p < 0.05 (two-tailed). All statistical analyses were conducted in R 4.3.1.
Results
Sexual dimorphism in morphological traits
Six of the 18 morphological traits exhibited significant differences between the sexes (Figure 2a), with males being consistently larger than females. Body length (BL; males = 8.188 mm, females = 7.749 mm) and body size (BS; males = 6.926 mm, females = 6.696 mm) showed the most pronounced differences (both p < 0.001), followed by pronotum dimensions (PronL: males = 3.254 mm, females = 3.057 mm; PronW: males = 4.352 mm, females = 4.064 mm), elytron width (ElyW: males = 4.350 mm, females = 4.177 mm), and protibia length (ProTiL: males = 2.060 mm, females = 1.909 mm) (all p < 0.01).
Sex differences in significant morphological traits and correlations between environmental factors and mean morphological traits in Onthophagus atripennis. (a) Morphological traits showing significant differences between females and males. Horizontal lines represent medians, boxes represent interquartile ranges, whiskers represent data ranges, and points represent individual observations. Females are shown in green and males in red. Only traits with statistically significant sex effects are shown. (b) Heatmaps showing correlations between environmental factors and mean morphological traits in females and males. Colours intensity indicates correlation coefficients ranging from −1.0 (blue, negative) to 1.0 (red, positive). Significance levels: *p < 0.05, ** p < 0.01, *** p < 0.001. Only environmental factors and traits involved in significant correlations are shown. All correlations are based on plot-level mean values (n = 7 plots per sex); due to the small sample size, these results should be considered exploratory.
Pearson’s correlation analysis revealed markedly divergent environmental sensitivity between the sexes (Figure 2b). Females exhibited coordinated responses to microhabitat quality: herb species richness (HS) correlated strongly with head dimensions (HeadL: r = 0.847; HeadW: r = 0.883), pronotum width (r = 0.955), and protibia length (r = 0.955), whereas tree height (TH) correlated positively with relative pronotum size (r = 0.929) and negatively with elytron length-to-width ratio (r = –0.964). By contrast, males showed more dispersed responses to macrohabitat gradients. Notably, HS correlated negatively with male body size (r = –0.793) – diametrically opposite to female patterns – whereas altitude (Alt) correlated positively with leg investment (ProTiL: r = 0.679; MetaTiL: r = 0.714; PronL/BS: r = 0.750) and negatively with mesotibia width (r = –0.786). Additionally, litter coverage (LC) was positively associated with greater male body length and pronotum width (BL: r = 0.883; PronW: r = 0.847), and shrub abundance (SA) correlated with enhanced flight capacity (ElyL: r = 0.893).
The most striking sex difference was the diametrically opposite effects of herb species richness (HS) on body size (BS): no significant correlation was detected in females, whereas a significant negative correlation was found in males. Furthermore, altitude (Alt) and litter coverage (LC) significantly affected only male morphology, whereas tree height (TH) affected only females.
Sex-specific patterns of phenotypic integration
Trait network parameters showed no significant differences between sexes (all p > 0.05; Figure 3a), with females exhibiting slightly higher modularity (Q: males = 0.068, females = 0.074) and average path length (AL: males = 1.325, females = 1.347), and males showing slightly higher network density (GD: males = 0.676, females = 0.659), though none of these trends reached statistical significance.
Sex differences in trait network structural parameters and correlations between environmental factors and trait network structural parameters in Onthophagus atripennis. (a) Comparisons of trait network structural parameters between females and males, including overall integration (GD), average path length (AL), and modularity (Q). Raw p values are shown in the plots. (b) Heatmaps showing correlations between environmental factors and trait network structural parameters in females and males. Colours intensity indicates correlation coefficients ranging from –1.0 (blue, negative) to 1.0 (red, positive). Significance levels: *p < 0.05, **p < 0.01, ***p < 0.001. Only environmental factors involved in significant correlations are shown. Correlations are based on n = 7 plots per sex; statistical significance in small samples may be fragile, and patterns require validation in larger datasets.
Pearson’s correlation analysis revealed divergent, sex-specific associations between network topology and environmental drivers (Figure 3b). In females, network integration responded positively to microhabitat structure: litter coverage (LC) correlated strongly with both increased network density (GD, r = 0.79, p < 0.05) and decreased modularity (Q, r = –0.71, p < 0.05); additionally, slope (Slo) also enhanced integration (GD, r = 0.76, p < 0.05). By contrast, female networks showed no association with altitude (Alt). Conversely, male networks responded exclusively to macro-scale gradients: altitude (Alt) correlated negatively with integration (GD, r = –0.85, p < 0.05) and positively with path length (AL, r = 0.89, p < 0.05), whereas males showed no significant response to litter coverage or slope.
Discussion
This study integrated single-trait and multi-trait network approaches to investigate both sexual morphological differentiation and sex-specific environmental responses in the dung beetle Onthophagus atripennis. Our findings suggest that sex differences in this species are expressed in mean trait values, in the response of phenotypic integration to environmental gradients, and in sex-specific environmental sensitivity patterns. Nevertheless, static trait network topology exhibited no significant sex differences. These results suggest that sex-dependent morphological adjustment represents a key mechanism for coping with environmental change in tropical and subtropical forests.
Sexual dimorphism and sex-specific environmental sensitivity at the single-trait level
Our results showed that male and female Onthophagus atripennis differed significantly across six traits, with differences concentrated in body size and functionally related structures, indicating clear trait specificity in sexual dimorphism (Blanckenhorn Reference Blanckenhorn2005; Stillwell et al. Reference Stillwell, Blanckenhorn, Teder, Davidowitz and Fox2010). Morphological dimorphic traits are consistently larger in males, a pattern consistent with stronger sexual selection on male competitive performance. (Stillwell et al. Reference Stillwell, Blanckenhorn, Teder, Davidowitz and Fox2010). Specifically, enlarged pronotum dimensions in males probably reflect investment in combat structures, given that competitive outcomes in male dung beetle depend critically on body strength and mass (Huerta et al. Reference Huerta, Cruz-Rosales, González-Vainer, Chamorro-Florescano, Rivera and Favila2023; McCullough and Simmons Reference McCullough and Simmons2016). Similarly, elongated protibiae in males may enhance digging, pushing, and substrate-manipulation performance during resource competition, as inferred from functional studies of dung beetles (Bijma et al. Reference Bijma, Gorges, Kovalev and Gorb2025). However, these morphological differences could alternatively arise from developmental plasticity or demographic sorting rather than pure adaptive optimisation. For instance, larger males may merely experience higher survival in competitive environments, creating observed patterns without implying functional coordination of specific trait modules (Tomkins et al. Reference Tomkins, Kotiaho and LeBas2005).
Previous studies on non-tropical Onthophagus species, particularly Onthophagus taurus, provide a valuable comparative framework for interpreting our findings. In Onthophagus taurus, male sexual dimorphism arises from resource-dependent developmental plasticity and alternative reproductive tactics: males that exceed a threshold body size develop enlarged horns and typically engage in fighting, whereas smaller, hornless males may adopt sneaking tactics (Moczek Reference Moczek1998; Moczek and Emlen Reference Moczek and Emlen2000). Moreover, studies on Onthophagus taurus have shown that larval nutritional conditions, population-level climatic gradients, and evolved plasticity can jointly shape adult body size, horn expression, wing morphology, and tibial morphology (Moczek Reference Moczek1998; Rohner et al. Reference Rohner, Macagno and Moczek2020). These findings are consistent with our observation that sex-dependent trait variation in Onthophagus atripennis reflects not simply fixed sexual dimorphism but may also be influenced by environmental filtering and developmental or demographic processes. However, unlike Onthophagus taurus, in which male horn polyphenism has been extensively studied, Onthophagus atripennis in our dataset lacks an obvious horn-based male morph system. Therefore, our results extend the discussion of sex-dependent morphological differentiation away from conspicuous weapon traits toward subtler whole-body and appendage traits, including pronotum size, elytral dimensions, and tibial traits.
Pearson’s correlation analysis revealed strikingly divergent environmental sensitivity between the sexes. Females exhibited coordinated responses to microhabitat quality (herb species richness correlated strongly with head dimensions, pronotum width, and protibia length), suggesting that female phenotypes are sensitive to local vegetation complexity (de Oliveira Ribeiro et al. Reference de Oliveira Ribeiro, Camargo and Frizzas2024). This finding indicates that environmental filtering may act not solely on single traits but rather on integrated phenotypes composed of multiple structures (Armbruster et al. Reference Armbruster, Pélabon, Bolstad and Hansen2014). The synchronous increase in multiple traits related to the head, pronotum, and forelegs in females from plots with higher herb species richness suggest that microenvironmental conditions associated with a more complex herb layer are linked to coordinated multi-trait responses.
By contrast, males showed dispersed responses across macrohabitat gradients. Multiple environmental factors each significantly correlated with different morphological traits, suggesting that male mean trait values are not governed primarily by a single environmental axis but rather jointly regulated by forest floor structure, vegetation stratification, and topographic gradients (Bui et al. Reference Bui, Ziegler and Bonkowski2020; deCastro-Arrazola et al. Reference deCastro-Arrazola, Andrew, Berg, Curtsdotter, Lumaret, Menéndez, Moretti, Nervo, Nichols, Sánchez-Piñero, Santos, Sheldon, Slade and Hortal2023; Nichols et al. Reference Nichols, Uriarte, Bunker, Favila, Slade, Vulinec, Larsen, Vaz-de-Mello, Louzada, Naeem and Spector2013; Salomão et al. Reference Salomão, González-Tokman, Dáttilo, López-Acosta and Favila2018). Notably, body size correlated negatively with herb richness – a pattern diametrically opposite to that observed in females – whereas altitude and shrub abundance predicted flight and dispersal-related traits. These patterns suggest sex-specific resource partitioning: females may be constrained by parental investment to respond to fine-scale microhabitat conditions, whereas males engage in broader mate-searching across landscape gradients (Schoener Reference Schoener1969). Critically, these findings extend previous work showing that dung beetle community recovery following disturbance is often initially male-biased by demonstrating that within-species sex differences in habitat sensitivity may underlie such demographic asymmetries (deCastro-Arrazola et al. Reference deCastro-Arrazola, Andrew, Berg, Curtsdotter, Lumaret, Menéndez, Moretti, Nervo, Nichols, Sánchez-Piñero, Santos, Sheldon, Slade and Hortal2023; Nichols et al. Reference Nichols, Uriarte, Bunker, Favila, Slade, Vulinec, Larsen, Vaz-de-Mello, Louzada, Naeem and Spector2013).
Sex-specific patterns of phenotypic integration and environmental response
While our limited sample size for network comparisons (n = 7) precludes definitive conclusions about static topological differences between sexes, these divergent environment–network associations provide preliminary evidence that sex-dependent differentiation may manifest more clearly in environmentally responsive trait covariation than in fixed network topology. These strikingly divergent environmental response patterns further suggest that sexual differentiation, where it exists, may be more strongly expressed in phenotypic differentiation than in fixed network properties. This interpretation aligns with the view that integration and modularity arise from coordinated versus independent trait variation and that environmental heterogeneity can reshape covariance architecture (Armbruster et al. Reference Armbruster, Pélabon, Bolstad and Hansen2014; Friedman et al. Reference Friedman, Remeš and Economo2019; Klingenberg Reference Klingenberg2008).
In females, litter coverage positively associated with overall integration and negatively with modularity, whereas slope also showed a positive association with integration. These patterns suggest that under thicker litter and steeper conditions, female trait networks become more coordinated and less modular. Such a shift implies that multiple traits may be adjusted in concert rather than through relatively independent modules, potentially facilitating efficient digging and brood provisioning (Armbruster et al. Reference Armbruster, Pélabon, Bolstad and Hansen2014). However, we caution that this pattern could alternatively reflect phenotypic differentiation in response to resource quality or demographic sorting of pre-adapted individuals, rather than adaptive coordination per se (de Oliveira Ribeiro et al. Reference de Oliveira Ribeiro, Camargo and Frizzas2024; Rohner et al. Reference Rohner, Macagno and Moczek2020).
Conversely, male networks showed reduced integration and increased path length at higher elevations, suggesting disrupted trait coordination under stressful macro-environmental conditions. This pattern may reflect the independent adjustment of flight and combat modules across elevational gradients (Nunes et al. Reference Nunes, Braga, Figueira, Neves and Fernandes2016; Stanbrook et al. Reference Stanbrook, Harris, Wheater and Jones2021). Because elevation integrates multiple concurrent environmental shifts including temperature, moisture, vegetation structure, and resource availability, this pattern suggests that male network structure is more responsive to broad-scale rather than to local microhabitat features. However, distinguishing phenotypic differentiation from genetic differentiation among separated populations requires reciprocal transplant or common garden experiments (e.g. Leeson et al. Reference Leeson, Kennington, Evans and Simmons2020).
Practical implications for forest restoration
Our findings suggest that effective restoration of dung beetle-mediated functions in tropical and subtropical secondary forests demands dual-scale monitoring that moves beyond species presence to secure sex-balanced population persistence. First, microhabitat assessment should target herb-layer diversity and litter coverage to guarantee female reproductive habitat quality. Second, landscape connectivity must preserve elevational corridors to facilitate male dispersal. Whereas current restoration practices typically emphasise canopy closure and tree diversity alone, our results reveal the critical importance of understory herbaceous communities and topographic connectivity for sustaining sex-balanced populations and overall functional recovery (Menéndez et al. Reference Menéndez, Monteith, van Oosterzee and Preece2024; Pozo-Rivera et al. Reference Pozo-Rivera, Quiloango-Chimarro, Paredes, Landívar, Chiriboga, Hidalgo, García and Villacís2023; Sullivan et al. Reference Sullivan, Slade, Bai, Shi and Riordan2018).
Limitations and future directions
This study has three key limitations. First, the research was restricted to a single region with only seven valid plots for network analyses, severely limiting statistical power and generalizability. Thus, our trait- and network-environment correlations must be interpreted as exploratory hypotheses requiring validation, not as established causal relationships. Second, sampling was limited to August–December 2021 owing to COVID-19 restrictions, resulting in a lack of seasonal dynamics and preventing assessment of reproductive investment effects (Akamine Reference Akamine2019). Third, the observational nature of the study prevents us from disentangling phenotypic plasticity, genetic differentiation, and demographic sorting. Resolving these mechanisms will require experimental or longitudinal approaches. Future research should include multi-season sampling, experimental designs, and broader spatial replication across environmental gradients. Comparative analyses across multiple dung beetle species, particularly within Onthophagus, would help establish whether the observed sex-dependent patterns are species-specific or generalisable across taxa. Similarly, expanding trait network approaches across diverse climatic regions would enhance both the robustness and generalizability of functional trait frameworks for intraspecific variation in dung beetles.
Conclusion
By integrating single-trait and trait network approaches, this study investigated sex-dependent morphological variation and its environmental associations in Onthophagus atripennis. Given the limited number of plots with sufficient sample sizes (n = 7), the following conclusions remain preliminary and require validation in larger-scale studies. Our results show that sexual dimorphism is not uniformly distributed across traits but is instead concentrated in body size and functionally relevant forebody and appendage structures, all of which are larger in males. Despite these mean-level differences, trait network structural parameters did not differ significantly between sexes, suggesting broadly similar static network topology within the studied population. However, this similarity in static network topology does not preclude sex differences in how network parameters respond to environmental gradients. Indeed, the relationship between morphology and environment revealed pronounced sex-specific patterns. Female mean trait values responded consistently to herb-layer structure, whereas male trait values were shaped by multiple environmental dimensions, including forest floor conditions, vegetation stratification, and topographic gradients. These sex-specific patterns extended to network-level responses. Female trait networks became more integrated and less modular with increasing litter cover and slope, whereas male networks exhibited reduced integration and longer path lengths along elevational gradients. Owing to the limited number of plots (n = 7), these findings remain exploratory. Within this context, sex differences in O. atripennis appear to be more strongly reflected in trait values and environment-associated variation than in intrinsic network structure. Incorporating sex-specific perspectives may therefore improve our understanding of intraspecific functional variation in soil fauna. In addition, this work contributes to a growing body of studies applying trait network approaches to intraspecific and sex-related variation in animals. The observed sex-specific habitat associations may provide initial ecological information to inform forest restoration strategies aimed at supporting soil fauna functional recovery in tropical and subtropical ecosystems.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0266467426100765.
Data availability statement
The data supporting the findings of this study are available within the article and its supplementary materials. Raw data are available from the corresponding author upon reasonable request.
Acknowledgements
We are grateful to Professor Ming Bai (Institute of Zoology, Chinese Academy of Sciences) for his insightful advice on bait formulation, and to Lu Qiu (Mianyang Teachers’ College) for his valuable assistance with morphological identification of the specimens.
Author contribution
Conceptualisation: Hui Bao, Fuling Yang, Shengbin Chen; Analysis: Hui Bao, Fuling Yang, Kunming Zhao; Investigation: Hui Bao, Fuling Yang, Shuanghui Huang, Yanru Jia, BA; Writing: Hui Bao, Shengbin Chen; Supervision: Shengbin Chen; Project administration: Shengbin Chen; Revision: all authors.
Financial support
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Competing interests
The authors declare none.
Ethical standards
The authors assert that all procedures contributing to this work comply with applicable national and institutional ethical guidelines. No specific permits were required for the field sampling described in this study.
AI declaration
During the preparation of this work, the authors used Kimi for language polishing of the manuscript. The tool was not used for data analysis, interpretation, or the generation of scientific content. The authors reviewed and edited all content and take full responsibility for the integrity of this publication.
