How to Become a Generalist Species? Individual Niche Variation Across Habitat Transformation Gradients

Species in transformed habitats, frequently labeled as environmental generalists, tend to show broader niches than species in natural habitats. However, how population niche expansion translates into changes in the niches of individual organisms remains unclear, particularly in the context of habitat transformation. Niche expansion could be a product of individuals having broader niches, greater distances among individuals’ niches, or a combination of both processes. This would challenge the traditional conceptions on niche dynamics, which emphasize the role played by individual specialization (IS). Here, using stable isotopes, we computed total niche width (TNW), its within- and between-individual components (WIC and BIC), and IS (the ratio WIC/TNW), in 13 populations of 6 bird species and 8 populations of 3 frog species in natural and transformed habitats. We confirmed that species had broader niche width in transformed than in natural habitats, yet population niche expansion across habitats was mainly a product of increased distance between individuals. Within each habitat type, increases in TNW were linked to increases in WIC for all habitat types, while relationships between TNW and BIC were found in transformed but not in natural habitats. Hence, both increased individual niche width and increased distance among individuals were apparent within habitats, particularly in transformed ones, where increases in WIC dominated. Neither across or within habitats was niche expansion associated with increasing IS. Therefore, our results overturn traditional conceptions associated with the niche variation hypothesis and illustrate that niche expansion is not invariably associated with increased IS, because the distance between individual’s niches (BIC) can increase, as well as the breadth of those niches (WIC).


INTRODUCTION
Habitat transformation is one of the main drivers of biodiversity loss (Newbold et al., 2015). This loss is not neutral in that predominantly specialized species -those requiring particular habitat or food resources -are disproportionately lost, while environmental generalists -those with less selective habitat or dietary needs -become dominant (Flynn et al., 2009;Ducatez et al., 2018;Sol et al., 2020). A consequence of this process is that species in transformed habitats display broader niches than species in natural habitats (Bonier et al., 2007;Clavel et al., 2011;Coogan et al., 2018;Pagani-Núñez et al., 2019;Palacio, 2019). It is unclear however how this process of population niche expansion translates into the niches of the individual organisms.
Total niche width (TNW) of a population is composed by the sum of its within-individual component (WIC) and betweenindividual component (BIC) (Roughgarden, 1972(Roughgarden, , 1974, while individual specialization (IS) is computed as the ratio between WIC and TNW (Bolnick et al., 2003;Araújo et al., 2011), and increases as this ratio decreases. Traditionally, the niche variation hypothesis (NVH) predicts that TNW is positively correlated with IS (Van Valen, 1965), meaning that increases in TNW are mostly associated with increasing IS (Cachera et al., 2017;Maldonado et al., 2017). This is because when TNW increases, IS is likely to increase too, simply because TNW would increase at a higher rate than WIC. This seems a realistic assumption, as generalist populations are often composed of individuals only using a subset of the total population niche (Araújo et al., 2011;Layman et al., 2015;Pagani-Núñez et al., 2015).
Many studies, however, have cricitized and/or found no support to the NVH (e.g., Simberloff et al., 2000;Meiri et al., 2005). This suggests that population niche expansion could be achieved through different mechanisms than an increase in IS, such as changes in other components of individuals' niches or a population's niche structure. As stated above, when a population niche expands (TNW increases), the value of the ratio WIC/TNW decreases automatically, as long as WIC does not necessarily change, or at least that it increases less than TNW and thus the main increase is in BIC (because TNW = BIC + WIC). We suggest here that if TNW increases the components of TNW likely increase too, yet this process of niche expansion could be achieved in many different ways (either BIC, WIC or both may increase). It is thus fundamental to consider how changes in these different niche components shape patterns of niche expansion, and habitat transformation gradients offer an excellent scenario to test these ideas.
Increases in both WIC and BIC are likely to occur across transformation gradients. Assuming that population size stays constant, this process of niche expansion at the population level (i.e., increasing TNW) could be reflected at the individual level through three main avenues: increasing WIC, increasing BIC, or increasing both WIC and BIC (Figure 1). Several extrinsic and intrinsic factors could facilitate this process. For instance, habitat heterogeneity could promote greater BIC -less overlap among individuals (Darimont et al., 2009;Newsome et al., 2015). Conversely, innovativeness could promote greater WIC -larger individual niches (Sol et al., 2013;Ducatez et al., 2015). Furthermore, these processes may also be apparent within habitats -particularly transformed ones-as WIC and BIC may increase with increasing population niches.
In this study, we assessed the role played by different niche components (WIC, BIC and the ratio WIC/TNW) in explaining patterns of niche expansion across habitat transformation gradients. We did this using six bird and three frog species inhabiting natural, rural and urban habitats in China, which harbors tremendous biodiversity yet is currently experiencing a vast process of habitat transformation driven particularly by urbanization (Seto et al., 2011;Liu et al., 2018). Birds and frogs strongly respond to habitat transformation and human activities, and are commonly used as models to investigate this issues (McKinney, 2008;Herrera-Montes and Aide, 2011). Furthermore, there is a wealth of research using stable isotopes to answer different questions about these two taxa (e.g., Inger and Bearhop, 2008;Araújo et al., 2009;Smith et al., 2017). While birds and frogs certainly are different (frogs would be less mobile and to some extent more specialized than birds, at least in relation to habitat use, due to their dependency from water) (e.g., Dayananda et al., 2017), we confirmed the consistency of our hypotheses by pooling together data from both taxa. To compute niche characteristics we used stable isotopes, which has become a common technique to quantify trophic niche (Layman et al., 2012;Figgener et al., 2019). First, we asked if species had broader niches in transformed than in natural habitats, and determined whether differences in WIC, BIC and the ratio WIC/TNW existed across natural, rural and urban habitats. Second, we assessed the relationships between these niche components and TNW within habitats. In doing so, we were able to determine which of these individual niche components was most important in explaining population niche expansion (Figure 1).

Study Locations and Species
We passively sampled passerines from November 2015 to April 2017 using mist nets in six locations in Guangxi Zhuang Autonomous Region and Yunnan Province (southwest China), and frogs during July-August 2018, actively capturing them in six locations in Gansu and Shaanxi Provinces (northwest China) (Figure 2 and Supplementary Material 1). We categorized our study locations as urban, rural and natural habitats, according to land-use gradients and the level of anthropogenic disturbance (cf. Newbold et al., 2015). Natural habitats were relatively undisturbed and protected forested areas, rural habitats were predominantly agricultural areas or managed forests with low human population density, and urban habitats were humandominated areas inside cities (Liang et al., 2018;Pagani-Núñez et al., 2019; Figure 2). Six common passerines in the area included the Rufous-capped Babbler Stachyris ruficeps, the Gray-checked Fulvetta Alcippe morrisonia, the Common Tailorbird Orthotomus sutorius, the Red-whisked Bulbul Pycnonotus jocosus, the Redbilled Leiothrix Leiothrix lutea and the Scaly-breasted Munia Lonchura punctulata. Frogs belonged to three species of the genus Feirana (Swelled-vented Frog F. quadranus, Taihangshan Swelled-vented Frog F. taihangnica and Kangxian Swelled-vented Frog F. kangxianensis) (Figure 3; Fei et al., 2009;Yang et al., 2011). For birds, we sampled a large number of individuals and species (see Pagani-Núñez et al., 2019). We focused, however, on six common species that were present in several habitats and limited sample size to 10 individuals per species (Figure 3). We did this by randomly selecting up to 10 individuals per species for stable isotope analysis. In doing so, we were able to use a similar number of individuals per species to compute niche characteristics. For frogs, since sample size per species was rather similar but lower in total than for birds, we did not set any sample size threshold. In the end, we used 176 individuals (118 birds and 58 frogs) from 21 populations (Supplementary Material 2).

Sampling Procedures
After birds were extracted from the mist net, we measured body mass using an electronic balance with an accuracy of 0.01 g. We then collected the second secondary feather (P8) of the right wing and the claw tip of the largest claw from the right leg. All individuals were banded before being released and we only included data from the first capture.
We searched for frogs shortly after sunset in and around water bodies using flashlights. For each captured adult, we measured body mass to the nearest 0.01 g with an electronic balance. After euthanasia, we clipped the fourth toe for molecular identification and collagen extraction. We also collected the muscle tissue from hind limbs, washed the muscles with distilled water and stored them in 2 ml tubes. All the samples were kept in cold conditions in the field and then stored at −20 • C in the laboratory. After sampling, the adult frogs were placed in 4% buffered formalin and the voucher specimens were deposited in the Herpetological Museum of the Chengdu Institute of Biology (CIB), Chinese Academy of Sciences (CAS). These procedures are part of a long-term line of research focused on the ecology and evolution of Feirana frogs (e.g., Hu and Jiang, 2018;Wang et al., 2019;Huang et al., 2020), and are a common practice in other stable isotope studies on this taxon (e.g., Araújo et al., 2009). For frog species identification, we integratedly referred to morphological characteristics, and geographic and genetic information (Hu and Jiang, 2018;Wang et al., 2019;Huang et al., 2020).
All animal handling and processing were in accordance with the Law of the People's Republic of China on the Protection of Wildlife and approved by the Guangxi Forestry Bureau and the Animal Care Committee of CIB, CAS.

Tissues and Niches
Nitrogen stable isotope ratios (δ 15 N) are commonly used to estimate species' trophic level (Boecklen et al., 2011;Layman et al., 2012). We focused only on this element because we were particularly interested in linking our finding to individuals' and species diets. Using different tissues it is possible to determine individual specialization because different tissues reflect individuals' diets across different temporal scales (Vander Zanden et al., 2015;Bond et al., 2016;Maldonado et al., 2017). Thus, we were able to compute intra-individual niche variation using δ 15 N values of two easy-to-obtain tissues.
We collected feathers and claws (birds) and muscle and bone collagen samples (frogs) to compute niche characteristics because these tissues provide information on niche use at two different temporal scales for each individual. Feathers correspond to a specific temporal window during summer molt, while claws correspond to the previous weeks to months prior to sample collection (Bearhop et al., 2003;Hahn et al., 2014;Vander Zanden et al., 2015;Bond et al., 2016). We collected these samples continuously across our study period so that they represent diets across a broad temporal scale. Similarly, muscle tissue would reflect niche use in a time window of weeks prior to sample collection, while bone collagen tissue would reflect a broader time span across individuals' development (Vander Zanden et al., 2015;Matsubayashi et al., 2017).
Thus, we used δ 15 N values from two different tissues from each individual as the continuous data input about niche use to calculate TNW, WIC, BIC and WIC/TNW for each species per location. WIC/TNW ranges from 0 (individuals use completely different resources, i.e., display high IS) to 1 (individuals use the full range of the population niche, i.e., display low IS). In our sample, WIC ranged between 0.04 and 1.13, BIC between 0.05 and 1.62, and TNW between 0.15 and 2.53 (Supplementary Material 2).

Stable Isotope Analysis
Claws and feathers from birds were cleaned from surface contaminants using NaOH (0.25 mol/L), and then were airdried for at least 12 h. We included approximately 0.35 mg of feather tips and claws of each individual into separate tin cups, which were loaded for δ 15 N analyses. We used a Flash 2000 HT elemental analyzer coupled via ConFlo IV Universal Continuous Flow Interface (Thermo Scientific) to a Thermo Scientific DELTA V Advantage isotope ratio mass spectrometer (Thermo Fisher Scientific, Bremen, Germany). We conducted stable isotope analyses for this taxon in the Stable Isotope Laboratory of the College of Forestry at Guangxi University (Nanning, S China). Toe clips from frogs were demineralized for 24 h in 0.5 M HCl, rinsed twice with deionized water. Next, the samples were treated with 0.1 M NaOH for 12 h and washed twice with deionized water. We immersed the bones in a methanol:chloroform mixture (2:1, vol:vol) for about 24 h, then thoroughly rinsed and heated in deionized water at 90 • C for about 12 h. Bone collagen and muscle tissues from frogs were freeze-dried for 36 h to constant weight, grounded to a fine powder and analyzed at the Public Center of Experimental Technology in CIB, CAS. We used a vario isotope cube elemental analyzer (vario ISOTOPE cube, Elementar, Germany) interfaced to an isotope ratio mass spectrometer (IsoPrime100, Isoprime, United Kingdom).
For both taxa, stable isotope ratios were expressed as parts per thousand ( ) using the equation as follow: δ 15 N ( ) = [(Rsample/Rstandard)-1] × 1000, where Rsample and Rstandard are the 15 N /14 N ratios of sample and standard, respectively. The standard was atmospheric nitrogen (AIR N 2 ) and measurement precision was 0.25%.

Statistical Analysis
Using a linear mixed-effects model, we first tested whether TNW was larger in rural and urban habitats than in natural ones. TNW was our response variable, and habitat type (natural, rural or urban), taxa (bird or frog) and species' average body mass were our predictors. We included body mass as it may have played a relevant role in shaping niche community structure and due to the differences between the studied taxa (Vergnon et al., 2009). Study location (Supplementary Material 1) and species (Figure 3 and Supplementary Material 2) were the random factors. Using the same statistical approach, we were interested in determining whether differences in BIC, WIC and IS (the ratio WIC/TNW) existed among habitat types (in parallel to differences in TNW). Thus, we constructed three different models with the same predictors and random variables yet subsequently using BIC, WIC and IS as response variables.
Second, we were interested in assessing the relationships between the different niche components (BIC, WIC and IS) and TNW across habitat types. To do this, we constructed a set of models subsequently using niche characteristics as response variables (BIC, WIC and IS). Taxa, body mass and the interaction between TNW and habitat type were the predictors, and study  location and species were the random factors. Habitat type factor had three levels (natural, rural and urban), so that we dummy coded these factors and repeated the analysis using a different level as reference to ascertain all the possible combinations (e.g., natural vs. rural, rural vs. urban and natural vs. urban). Furthermore, there was considerable altitudinal variation among locations, and in spite that we have previously recorded negligible altitudinal effects on species' niche characteristics (Pagani-Núñez et al., 2019), we assessed whether it impacted individual level metrics. We recorded no significant effects of altitude on niche characteristics, so we excluded this variable from the final models (TNW: P = 0.37, WIC: 0.11, BIC: 0.81, IS: 0.21, data not shown). We visually examined qqplots to confirm that model residuals were normally distributed. From the first set of models assessing differences in niche characteristics across habitat types, we found that the one using WIC as dependent variable did not fit this criterion. Thus, we repeated this model using log-transformed WIC values to approximate normality. We computed niche characteristics using the function WTcMC from the RInSp package (Zaccarelli et al., 2013) and constructed the linear mixed-effects models using the package lme4 v1.1-23 (Bates et al., 2015). All analyses were performed in R software v3.6.1 (R Core Team, 2017).

Differences in TNW, WIC, BIC, and IS Across Habitat Types
We found that frog and bird urban populations had larger TNWs than natural populations ( Figure 4A and Table 1). Urban populations showed a non-significant tendency to display larger TNW than rural populations, while rural and natural populations showed no apparent differences in TNW (Table 1). We also found that TNW correlated negatively with body mass. Bird and frog populations did not show significant differences in WIC across habitat types ( Figure 4B and Table 1). WIC and body mass were uncorrelated ( Table 1).
Between-individual component of bird and frog urban populations showed a non-significant tendency to be larger than that of natural populations ( Figure 4C and Table 1), while rural populations did not show significant differences with natural or urban populations. BIC and body mass were uncorrelated (Table 1).
We did not record any significant effect of habitat type or body mass on IS (WIC/TNW) ( Table 1). Taxa showed no significant effects on niche characteristics (Table 1).

Relationships Between TNW and WIC, BIC and IS Within Habitat Types
Within-individual component and TNW correlated positively in natural, rural and urban populations, the former showing the steepest curve ( Figure 4D and Table 2). BIC showed a different pattern. BIC and TNW showed a strong positive correlation in both urban and rural populations but not in natural populations ( Figure 4E). Finally, IS and TNW were uncorrelated across habitats ( Table 2). Taxa and body mass showed no significant effects on these models ( Table 2).

DISCUSSION
In this study, we illustrate how increases in population niche width (TNW) in birds and frogs showing different patterns across and within habitats. Within habitats, increases in TNW were a product of both increased niches within individuals (WIC), namely increased individual niche width, and increased distance between individuals (BIC), namely reduced overlap among individuals. These processes occurred simultaneously, thus, supporting our third hypothesis on individual niche dynamics across habitat transformation gradients (Figure 1). These positive relationships between TNW, WIC and BIC were rather persistent within habitats, with the exception of the relationship between TNW and BIC, which was not significant in natural habitats. Conversely, across habitats, although TNW increased markedly, we recorded no differences in WIC and only a weak tendency of BIC to increase from natural to urban habitats. Individual specialization (IS) showed no significant relationships with TNW across or within habitats. Hence, our results highlight how increases in TNW can be achieved by a combination of individual and population level niche changes, and is not invariably associated with an increase in IS.

Differences in Niche Characteristics Across Habitat Types
We found that species' niche expansion from natural to transformed habitats, namely increased TNW, did not necessarily imply dramatic changes in both WIC and BIC. While we did record markedly larger TNW in urban than in natural habitats, differences in WIC were negligible and species only showed a non-significant tendency to have larger BIC between these two habitat types. Many studies using different approaches have reported similar patterns of niche expansion across habitat transformation gradients (Bonier et al., 2007;Devictor et al., 2008a;Callaghan et al., 2019;Pagani-Núñez et al., 2019;Palacio, 2019), yet no study had thoroughly addressed how this pattern is a result of changes in different niche characteristics at the individual level. Our study is original in showing this pattern of trophic specialization within species, while previous studies analyzing changes in specialization mostly focused on interspecific comparisons in habitat use (Julliard et al., 2006;Devictor et al., 2008b;Rivas-Salvador et al., 2019). Nevertheless, having a clear idea of what happens to individuals is fundamental to avoid biased estimations of species-or population-level estimation of isotopic niches (Flaherty and Ben-David, 2010). Larger TNWs could be a product of urban individuals showing high behavioral innovativeness typical of generalist species (Ducatez et al., 2015;Sol et al., 2016;Griffin et al., 2017), having broad available empty niches (Sanz-Aguilar et al., 2015;Yurkowski et al., 2016;Balme et al., 2019), or a combination of both processes. However, since BIC increased more than WIC across the transformation gradients, our results suggest therefore that habitat heterogeneity would be the main facilitator of niche expansion -promoting lower overlap among individuals within each species (Darimont et al., 2009;Newsome et al., 2015). We cannot discard that changes in aggressiveness toward conspecifics across habitats may have influenced BIC values too.

Differences in the Relationships Between Niche Characteristics Within Habitat Types
We recorded different relationships among niche characteristics within habitat types. A positive relationship between TNW and WIC was apparent in natural, rural and urban habitats. We also recorded strong positive relationships between TNW and BIC in rural and urban habitats, but not in natural habitats. Both WIC and BIC facilitated niche expansion (Figure 1), and our results showed that increased distance among individuals, i.e., BICs or decreased inter-individual niche overlap, was only possible in transformed habitats, where population niches (TNWs) were large enough to allow this process to occur, supporting the idea that increasing habitat heterogeneity may have facilitated niche expansion across the transformation gradient.
In contrast with previous studies using a diverse array of statistical approaches Navarro et al., 2017;Murray et al., 2018), we recorded no significant relationships between TNW expansion and individual specialization (IS). Traditionally, the niche variation hypothesis suggests that increases in niche width (TNW in our case) are mainly achieved as a main consequence of a parallel increase in IS (Ebenman and Nilsson, 1982;Dayan and Simberloff, 1994;Bolnick et al., 2007;Cachera et al., 2017;Maldonado et al., 2017). Conversely, our study suggests that WIC, and especially BIC, are predominantly associated with increases in TNW. We recorded this pattern using two rather different taxonomic groups, birds and frogs, which suggests that this fashion may be common throughout vertebrates.

CONCLUSION
Generally, research investigating the consequences of habitat transformation for niche characteristics of species and populations (and, therefore, community assemblage dynamics) focus on broad scale processes (Darimont et al., 2009;Quevedo et al., 2009;Newbold et al., 2018;Sol et al., 2020). Thus, it remains unclear how individuals respond to these changes at finer scales. In other words, environmental generalists have broad niches and are able to colonize many habitat types but, so far, it was unclear how their individual niches were configured (Ducatez et al., 2018). Our research is important in linking populationlevel niche processes to individual-level niche characteristics. Although we may acknowledge that our sample size was rather limited and we were pooling different taxa (birds and frogs), our findings overturn traditional conceptions on the mechanisms and consequences of niche expansion, illustrating diverse ways in which it can be achieved. Further research could ascertain to what extent individual-level niche characteristics of transformed habitats are a consequence of individuals being generally more innovative and thus exploiting a broad variety of food and habitat resources or whether they merely reflect a particular habitat structure that enables individuals having broad niches without overlapping with conspecifics (Ducatez et al., 2015).

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: NCBI (https://www. ncbi.nlm.nih.gov/) MT940544-MT940569 (26 16S sequences) and MT941534-MT941565 (32 ND2 sequences).

ETHICS STATEMENT
The animal study was reviewed and approved by Animal Care Committee of CIB, CAS and Guangxi Forestry Bureau.

AUTHOR CONTRIBUTIONS
EP-N, YL, JH, DL, SY, and EG conceived and designed the study. EP-N, CH, DL, and SY performed the study and collected the data with assistance from YL, JH, and WL. EP-N, SY, and DL analyzed the data and wrote the draft of the manuscript with input from all authors. grateful to the Forestry Bureau of Guangxi Zhuang Autonomous Region, to the managers of Damingshan National Reserve, Gaoligongshan National Reserve, Longshan Regional Reserve, Medicinal Botanical Garden of Nanning and Gaofeng National Forest Park for providing permissions.