Females and Males Respond Differently to Urbanization (the Case Study in Ground Beetles)

1. Introduction

Land-use change is considered the primary source of biodiversity loss worldwide [1], and urbanization represents one of the most intense and long-lasting modifications of natural systems [2]. Urbanization gradients, typically extending from rural areas to highly developed city centers, have long been a major target of ecologists' research. Such gradients include extensive changes in habitat structure, including habitat loss and fragmentation [3] and increased soil impermeability, leading to the subsequent urban heat island effect.

With all this diversity of changes in the urban environment, it is not surprising that the response of the biotic component to urbanization can be diametrically opposed. For example, it has been shown that some butterfly species are completely absent from cities, although they inhabit adjacent areas [4], while the abundance of some midge and spider mite species in urban ecosystems increases [5].

As for ground beetles, a meta-analysis based on publications from 14 European and American cities showed that it is not so much the number of beetles that changes in cities, but rather the structure of communities. Changes occur in size composition (in favor of small species), flight ability (in favor of macropterans and dimorphic species), and trophic preference (in favor of herbivorous species and species with mixed feeding) [6]. At the same time, these authors note that despite the close relationship with these parameters, other sources of variability in urban ground beetle communities must be considered, as no correlation with either city size or its history (date of foundation) has been established, and some ground beetle species demonstrate diametrically opposed spatial exploration strategies in the cities studied.

Anthropogenic transformation in invertebrates leads to increased plasticity of physiological traits, while plasticity of morphological traits decreases. Anthropogenic influence is more detrimental to animal populations compared to fluctuations in the natural environment, since it leads to more rapid restructuring [7]. Thus, changes in the external environment in general, and anthropogenic influence in particular, can lead to maladaptation of populations. To avoid extinction, populations develop responses (increased fitness, population size, etc.). Recently, increasing attention in this regard has been paid to phenotypic plasticity [8,9], which is defined as changes in the phenotype of an organism under the influence of the environment. In some cases, such plasticity is adaptive and allows a population to “jump” from one peak of fitness to another, by passing the “passage” of fitness valleys [8,10].

Morphological features make it possible to expand the scope ecological-functional responses of a guild (in this case, ground beetles) from local to global scales, thus bypassing a purely taxonomic approach based on a regional species pool [11]. A similar concept from local to global is outlined by Pizzolotto et al. [12], but only for chorological traits, where a new index of biogeographic specificity is proposed that emphasizes the value of endemic species for different types of biotopes. The use of species traits in terrestrial habitats provides a new opportunity to assess the degree of disturbance of a biotope if the ecology, morphology, physiology, and life cycles of a taxon are well studied.

Morphology is the main phenotype of an organism, directly related to how it interacts with the environment [13]. Therefore, the study of the morphological characteristics of species, their diversity, and variability complements knowledge about populations and communities, provides insight into factors influencing organisms' responses to environmental changes.

Our research was inspired by a review publication by Estonian researchers, where sexual dimorphism was assessed based on the degree of variability in the sizes of females and males, which, according to the authors, reflects the sensitivity of the sexes to environmental factors [14]. To quantitatively describe the relationship between the sizes of females and males, we used type II regression (RMA II), which involves constructing a regression curve of the logarithm of male sizes on the logarithm of female sizes. The model constant (Intercept), slope, and statistical significance are also calculated. The results of the regression analysis were interpreted as follows. A positive slope of the regression curve indicated that environmental conditions influenced the sizes of females and males in the same direction, i.e., male size increased with increasing female size. This result allowed us to unambiguously rank the samples according to the degree of favorability of environmental conditions. Furthermore, a zero model constant implied a proportional increase in the sizes of males and females as conditions improved. A positive model constant indicated that females increased in size relatively faster than males, with males being more sensitive to environmental conditions. A negative model constant indicated the opposite trend. These parameters allowed us to draw conclusions about the dependence of sexual dimorphism on environmental conditions. In particular, it is easy to see that sex differences in the sensitivity of body size to environmental conditions should lead to different size ratios of females and males in different environments. Accordingly, with increasing environmental quality, the greater sensitivity of female body size will lead to more pronounced sexual dimorphism in species with larger females. The latter includes the ground beetle species C. granulatus, which we studied. If the model constant is negative, the slope of the regression curve will be greater than 1, and, accordingly, males are more sensitive in this case. All conclusions about the dependence of sexual dimorphism on body size and environmental conditions were made at the meta-level. For this purpose, a vote-counting method was used (positive versus negative slopes; positive or negative model constants). The parameters of each intraspecific relationship were treated as separate observations.

In our study, we essentially replicated the methodology discussed above, but applied RMA II models specifically to the analysis of beetle size variability depending on collection location within a range. Carabus granulatus L. was chosen as the model species due to its widespread distribution, sufficient eurybionticity, and well-studied ecology. Similar studies have been conducted previously on other ground beetle species. Thus, in relation to the mountain ground beetle Carabus cummanus, different sensitivity of females and males to the vegetation conditions of the biotope was shown [15], and in other species of ground beetles - Carabus odoratus, Carabus exaratus, Pterostichus montanus - different patterns of variability of individual organs in females and males at different altitudes in the mountains [16]. In the present study, we aimed to determine the nature of variability in the sizes of beetles of both sexes depending on the habitat. The latter was assessed by the degree of urbanization - cities, suburbs, natural biotopes. At the same time, we predicted: (i) the direction of size variability should be the same for females and males at all captured points; (ii) in urban habitats, males will be more sensitive to environmental conditions, that is, their size will increase at a greater rate compared to females; (iii) accordingly, sexual dimorphism in urban habitats will be expressed to a lesser extent compared to suburbs and natural biotopes.

2. Materials and Methods

2.1. Object of Study

C. granulatus is a large beetle that measures 18-26 mm in length, with a bronze or copper upper surface, often with a green sheen, and sometimes bronze black (Figure 1). The femora is sometimes red. The pronotum is uniformly densely punctate. The antennae are long, reaching halfway along the elytra in males. The elytra are sinuate before the apex with elytral eripleura abruptly near the upper edge of sinuation and have a sharp sculpture consisting of ridges and chains of tubercles. The pits on the elytra are small and barely noticeable [17,18].

This species is hygrophilous and eurytopic with a transpalaearctic distribution throughout the Eurasian continent from the lowlands to the mountains. It inhabits riparian ecosystems or moist floodplain meadows with dense vegetation. The species has a life cycle; in the northern taiga, it is found only in meadows, while in floodplain forests, its occurrences are rare [19]. In more southern regions, this moisture-preferential species is observed in moist to wet forests, especially floodplain [20,21]. In general, it is a spring breeder without larvae, but with an obligatory adult diapause [21]. It hibernates, often gathering in groups, under the loose bark of trees. Both adults and larvae are predators. The larva moults three times before pupation and, in the laboratory, beetles have been successfully reared from eggs in 53 days.

2.1. Study Area and Trapping

This study is part of a larger project on the morphometric variability of the ground beetle C. granulatus. As part of this study, a large team of carabidologists has been collecting samples of this beetle for many years, which are then transported to the Institute of Ecology and Subsoil Use of the Academy of Sciences of the Republic of Tatarstan (Kazan, Russia) for morphometric analysis. A detailed description of the beetle collection sites and collection methods is provided in our recent publication, which examined the latitudinal gradient of size variability in this species of beetle [22]. Since then, we have added beetle samples from Italy, Czech Republic, some regions from Russia. Thus, the database with morphometric measurements of six linear traits of beetles has reached 10,172 individuals.

The sampling area included a vast territory in different provinces of Russia, and several sites in six European countries (Figure 2, Table 1).

2.2. Morphometric Data

All measurements were conducted using a unified methodology adopted by us, described in previously published articles [23,24,25]. Studied plots differed in some environmental parameters. But taking into account that our investigation was large-scale and remembering the huge number of sampling plots and measured individuals, we argue that our approach was the high-throughput one. The last became very popular the last decades, when the authors compiled data even not their own research and make conclusions on the given topic [26].

Measurements included six traits: elytra length, elytra width, pronotum length, pronotum width, head length and distance between eyes (Figure 3).

2.3. Statistical Analysis

We used the R statistical environment.

An ANOVA was conducted to assess the influence of environmental factors.

Specific SSD values ​​were calculated using the generally accepted Loveich and Gibbson formula: SSD = females/males-1.

To estimate the allometry of the PDR, type II regression models were used, using the reduced major axis method (RMA). The logarithm of the quantile distribution of the studied male traits was calculated as a linear function of the logarithms of the quantile distribution of these sizes in females. The model is output to a text file, Y = άXβ. Positive values ​​of the regression coefficient (β) indicate the same direction of change in the trait in males and females; in other words, as the trait value increases in females, the values ​​of the same trait in males also increase. Positive values ​​of the model constant (Intercept) indicate that females increase in size faster than males, meaning that females are more sensitive to environmental changes; negative values ​​indicate the opposite [14]. RMA and SMA should be used when there is a high correlation (we have this - on average about 75% and higher with a high level of reliability p<0.001 according to Pearson), and also when the number of points is more than 67 (we had that) [27]. The lmodel2 library was used to calculate the RMA regression coefficients [28]. For example

result_ <- lmodel2(el_l_male_log ~ el_l_female_log, data = data,

range.y = "relative",

range.x = "relative",

nperm = 1000)

print(result_)#(factor «range» was taken for estimation of Ranged Major Axis)

Regression results

      Method Intercept      Slope Angle (degrees)     P-perm (1-tailed)

1      OLS        3.914979 0.7519160         36.94009         0.000999001

2      MA        1.189834 0.9910264         44.74177         0.000999001

3      SMA        1.165074 0.9931990         44.80450        NA

4      RMA       1.397337 0.9728197         44.21066        0.000999001

Confidence intervals

       Method       2.5%-Intercept       97.5%-Intercept       2.5%-Slope       97.5%-Slope

1      OLS        2.0799836        5.749974        0.5911750        0.912657

2      MA        -1.5281748        3.384476        0.7984635        1.229511

3      SMA        -0.8141801        2.849755        0.8453812        1.166863

4      RMA        -1.2463576        3.567658        0.7823908        1.204783

Eigenvalues: 0.7612505 0.1052455

H statistic used for computing C.I. of MA: 0.01142403

To calculate the regression, we used logarithmic mean trait values ​​for males, based on the mean trait value for females from each sampling point. There were 67 sampling points in total, including 43 from natural habitats, 10 from suburbs, and 14 from cities.

The fragment of data frame used: locality anthropel_l_male_logel_l_female_log

<int><int><dbl><dbl>

1        1        1        2.56        2.42

2        2        1        2.56        2.44

3        3        1        2.56        2.42

4        4        1        2.55        2.42

5        5        1        2.56        2.44

6        6        2        2.54        2.41

7        7        2        2.52        2.41

8        8        3        2.53        2.48

9        9        3        2.52        2.48

10        10        3        2.55        2.42

3. Results

The ANOVA showed that the level of anthropogenic impact on ground beetle habitats influenced the size of their traits. Table 2 presents the results for the effect of anthropogenic pressure on elytra length, showing that the level of anthropogenic impact and sex, as well as their interaction, influence elytra size in the studied ground beetle species.

Roughly similar results were obtained for other traits (Appendix A, Table A1, Table A2, Table A3, Table A4, Table A5, Table A6, Table A7, Table A8, Table A9, Table A10 and Table A11), but there were some differences. For example, no statistically significant combined effects of the "Anthropogenic" and "Gender" factors or the effect of suburban habitat were observed for elytra width. No effects of urban and suburban habitat were observed for pronotal length. Variability in pronotum width depended solely on gender, and the interaction of the "Anthropogenic" and "Gender" factors affected head parameters.

Thus, it was demonstrated that not only the level of anthropogenic impact but also its interaction with gender influenced beetle size. The involvement of gender in regulating the biota's response to anthropogenic impact led to changes in SSD.

We first calculated the average SSD values ​​in plots with different levels of anthropogenic impact (Figure 4). To our surprise, the highest SSD values ​​were consistently recorded for all six studied traits in suburban areas: SSD values ​​for elytra and pronotal length, as well as distance between eyes, were statistically significantly higher in suburban populations compared to those from the city and natural habitats. SSD values ​​for elytra and pronotum width did not differ between beetles from natural habitats and suburban areas, and for head length, they were similar between beetles from suburban and urban areas.

To further our understanding of the mechanisms underlying this SSD variability, we analyzed the data using the RMAII method (Table 3).

The slope of the regression curve was positive in all cases, meaning that the sizes of females and males vary in the same direction depending on habitat. Generally, a pattern was observed: if the model constant was negative, the slope of the regression curve was greater than 1; if it was positive, it was less than 1. The two curves examined were an exception (12%).

If we consider each trait separately, we see that, under virtually all types of anthropogenic impact (cities, suburbs, and natural habitats), females exhibited greater sensitivity to environmental conditions. Regression curve slopes were greater than 1 in only three cases: elytra width in cities, pronotum width in suburbs, and interocular distance in natural habitats. In other words, urban habitat increased male sensitivity to the "elytra width" trait; suburban habitat increased sensitivity to the "pronotum width" trait; and in natural habitats, males were sensitive to the "interocular distance" trait. It should be noted that, regardless of the factor, body width traits were involved in the phenomenon of increased male sensitivity.

The final aspect of our research concerns the confirmation of the hypothetical figure presented in the article by Teder and Tammaru [14], whose work we relied on for our experiments. The figure presented by the aforementioned authors suggests that SSD should decrease with a negative Intercept and, accordingly, a regression slope greater than 1. This is observed because male size increases faster than female size, and in species with female-biased SSD.

We conducted a generalized RMAII analysis, which results in a logmale/logfemale regression curve for C. granulatus beetles as a whole for each studied trait. Specific beetle collection locations are indicated by colored circles in these same figures (Figure 5, Appendix B, Figures B1-B5).

Since each point marked in Figure 5 corresponded to a specific beetle sample (identified by a number in our database), we were able to calculate the SSD value for each sample to test whether SSD value indeed increases with beetle size and how this applies to samples from urban, suburban, and natural habitats.

To implement this hypothesis, we calculated the SSD value in each of the studied localities, arranging the data in the figures sequentially in order of increasing trait size. To illustrate this analysis, we present Figure 6, which presents SSD variability for a number of traits in C. granulatus populations in urban habitats.

For each curve, we calculated the R2 value of the trend line. Similar calculations were performed for C. granulatus populations inhabiting suburban and natural habitats, figures for which are presented in the Appendix B, Figures B1, B2.

Next, we analyzed the hypothesis that SSD values ​​should decrease with a negative Intercept of the models, and increase with a positive one [14]. In conducting our analysis, we relied on the same hypothetical assumptions [14] that SSD values ​​should decrease with a negative Intercept of the models, and increase with a positive one. Similar regression curves for other traits are shown in Appendix B, Figures B3-B7. We estimated the direction of actual changes in SSD values ​​with increasing trait size based on the slope of the regression curves for trait variability. The results are presented in Table 4, which shows that the assumption of directional variability in SSD with increasing trait value was not confirmed. Only in 8 cases out of 18 did the actual calculated SSD values ​​corresponded to their position on the regression curve.

4. Discussion

The topic of SSD in relation to ground beetles raised in this article is not new. In the past decade, a significant number of studies have appeared presenting data on the different patterns of size variability in female and male carabids. Sexual dimorphism in ground beetles is female-biased, and this has been confirmed across a fairly large number of species [29]. Latitudinal size variability has typically been studied, and the curves for male size variability have always been lower than those for females [23,25]. Statistically significant size differences were observed in the ground beetle Carabus aeruginosus between populations at different edges of its range, due in part to sexual differences [30]. In some regions, the sizes of females and males were comparable, but even for small ground beetle species, for example, in the genus Poecilus, the phenomenon persists that the curves of male size variation across a latitudinal gradient, although mirroring those of females, are significantly lower [31].

Similar sex-related size differences are also observed when analyzing ground beetle variability across an altitudinal gradient. In both Carabus and Pterostichus beetles, males are generally smaller than females at all altitudinal levels [32,33]. Moreover, many of the cited studies provide specific data on sexual dimorphism across both latitudinal and altitudinal gradients. The curve of sexual dimorphism variability along the latitudinal gradient is sawtooth-shaped, with SSD values ​​increasing or decreasing abruptly from one region to another. Even more interesting findings were observed along the altitudinal gradient. The curve of SSD variability is concave, with minimal values ​​in the mid-mountain regions, and correlates with the population sizes of the studied ground beetle species at these altitudes [16].

Regarding the species discussed in this article—C. granulatus—recently published studies have shown that the sizes of both females and males in this species tended to decrease across a latitudinal gradient, with the male variation curve also located at a lower level [22,24].

The topic of SSD was also addressed in these studies: its variation curve across a latitudinal gradient was sawtooth-shaped with a slight downward trend. Moreover, the highest SSD values ​​were observed for elytral length and pronotum width.

In this article, we are more interested in studies that estimate SSD values ​​based on RMAII results rather than directly calculated using the generally accepted formula [35]. The first such study on ground beetles was published quite a while ago [29]. The analysis of regression curves for 12 species presented there showed that their slopes were generally less than 1, indicating that the SSD in carabids was female-biased. This topic was subsequently raised repeatedly at conferences [e.g., 36, 37], where evidence was presented of variability in the position of the regression curve depending on the beetles' habitat—city, suburb, or natural habitat.

Our article also showed that urbanization influenced RMAII results. Looking at the RMAII results for all six traits (figures in Appendix B), we see that the regression curve plotted separately for biotopes with varying degrees of anthropogenic pressure (urban, suburban, natural) could change its position. However, this variability did not have a specific direction.

Our experimental data in no way detract from the merits of the Estonian authors, who were among the first to conduct a meta-analysis of SSD variability in insects. In recent decades, this approach has become increasingly popular because, as the saying goes, "big things can be seen from afar." In such publications, authors collect a large body of literature, extract factual data from it, compile information on a specific issue, and draw a general conclusion. Such publications are certainly welcome, if only because they address fundamental issues.

Some discrepancies between experimental data and hypothetical concepts of SSD variability are easily explained by logic.

Linear morphometric traits commonly used in field entomology do not correspond to simple theoretical expectations derived from body mass. The observed variability likely reflects complex, nonlinear response norms across multiple traits, rather than the uniform, monotonic response typically associated with biomass.

Given advances in computer technology, it is reasonable to assume that the use of RMAII cannot address the researchers' intended objectives. A more appropriate approach is the use of iteratively reweighted least square (IRLS) methods, which we have begun to explore.

Given that meta-analysis is increasingly becoming a practice for synthesizing data from different researchers and formulating general biological patterns, greater attention should be paid to large-scale studies conducted at the intraspecific level and using a unified methodology.

5. Conclusions

Our data demonstrate a non-monotonic response of SSD to the urbanization gradient. The peak values ​​in suburbs may indicate that moderate anthropogenic environmental transformation (the ecotone effect) creates conditions for the maximum realization of potential differences in growth and development between males and females. In turn, under conditions of severe stress (urban) or a stable environment (nature), these differences are neutralized.

Deviations from the general pattern were observed exclusively for traits related to body width. This suggests that these "volume" characteristics are the most sensitive indicator of male physiological state under anthropogenic stress, possibly due to differences in energy accumulation and expenditure strategies between the sexes.

Despite the lack of a universal trend across all traits, R² analysis revealed that the highest proportion of explained variance in SSD change with increasing body size was observed for elytra and pronotum length in urban biotopes, elytra width in suburbs, distance between eyes in natural biotopes. This suggests that different body parts obey different allometric rules in response to urbanization.