Butterfly community composition within a tropical urban landscape is influenced by habitat type and temperature

The specific factors that influence spatial community or population dynamics are often elusive, and even less known is the impact of tropical urban landscapes on diverse species community assemblages. To address this knowledge gap, we used a survey data set with 510 fruit‐feeding butterflies comprising 20 species across two heterogeneous habitats within a city in Nigeria. Next, we constructed generalised linear mixed models to understand the differential responses of the butterfly community to changes in environmental conditions across habitats. Butterfly species community assemblages significantly differed between the two urban habitats, with butterfly species significantly higher in the savannah woodland compared with the gallery forest due to the optimal daily temperatures of the savannah woodland. However, butterfly richness was lower in the gallery forest due to extreme environmental conditions. This study highlights that butterfly community changes in tropical urban landscapes are possibly responding to local microclimates and spatial heterogeneity across habitats. For evidence‐based conservation management of tropical butterfly biodiversity, there would be need for a long‐term, extensive and systematic insect monitoring programme for butterflies across disturbed and undisturbed fragmented habitats harbouring diverse insect species.


INTRODUCTION
Changes in environmental conditions play critical roles in structuring the community of species across ecosystems (e.g., Forister et al., 2018;Halsch et al., 2021;Hill et al., 2021;Nishio, 1980;Okoro et al., 2023;Parmesan, 2006).The influence of these environmental changes on community composition and structure can be on a global (e.g., increasing global climate change), regional (e.g., anthropogenic landscape changes) or local (e.g., changes in habitat types) scale.
Across heterogeneous urbanisation gradients in areas facing growing human influence (e.g., percentage of impervious surface), the environmental changes that influence adaptation strategies have been of great interest to community ecologists, and several studies have shown how urbanisation may be impacting global community composition, structure and species distributions (e.g., Diamond et al., 2022;Diamond et al., 2023;Perez & Diamond, 2019).However, despite over half of the world population living in urban areas (United Nations Department of Economic and Social Affairs, 2022), our understanding of what may be structuring or maintaining our local-scale biodiversity patterns in urban systems is limited (Pickett et al., 2017).Given our limited understanding of the factors structuring biodiversity in urban systems, the question of whether urbanisation tends to exacerbate species loss, as generally expected, is far from resolved (Hall et al., 2017;Menke et al., 2011;Saari et al., 2016).Thus, identifying specific environmental conditions driving local-scale community composition and structure across habitats in urban areas is important in informing urban biodiversity conservation.
Anthropogenic impacts in tropical urban systems, such as increasing land-use change, have continued to degrade key ecosystems (Shellberg et al., 2016), with tropical systems such as South America, southeast Asia and central Africa experiencing the most degradation (Baranowski & Bharti, 2023;Hansen et al., 2013).In the tropics, increasing ecosystem degradation with rising urbanisation rates is modifying ecosystem services and regional/local climates (Brenner et al., 2023;Fang et al., 2023;Foley et al., 2005;Okoro et al., 2023), having the potential to heavily impact habitat availability and resource use by native species (Baranowski & Bharti, 2023;Haddad et al., 2015;Oeser et al., 2023;Okoro et al., 2023).Rising environmental perturbation caused by land conversion is expected to impact many native urban species by modifying or removing their habitats and resources.Studies have shown that human-driven land conversion fragments the habitats of native species, facilitating invasion by exotics that changes community composition and structure (Brooks, 2003;Diamond et al., 2023;Shackelford et al., 2018;Tulloch et al., 2016;Woinarski et al., 2015).A study of butterfly diversity in southern Nigeria revealed a higher abundance and diversity index in protected relative to unprotected areas (Orimaye et al., 2016), indicating that human-related activities are negatively impacting the butterfly communities.Thus, urban landscapes experiencing increasing environmental changes may be facing significant declines in species richness, diversity and abundance (Crooks, 2002;Edwards et al., 2014;Oeser et al., 2023).Therefore, research on biodiversity conservation while promoting landscape sustainability in urban areas, primarily in tropical climates, is needed to improve ecosystem services in these landscapes (Diamond et al., 2023;Fang et al., 2023).Specifically, investigations on the impacts of urban expansion on biodiversity patterns in tropical areas of Africa with more developing economies are particularly needed, given that most urban growth in the near future is predicted to occur in this region (Dos Santos et al., 2017).
Among the species found in tropical, urban habitats, butterflies occupy a range of niche landscapes, allowing for wider global distributions.In addition to being appealing to the general public, butterflies can also be used as indicators of how environmental perturbation can influence local-scale communities because of their sensitivity to small environmental changes (Blair, 1999;Essens et al., 2017).For example, establishing a viable butterfly population may largely depend on the extent of habitat connectivity and/or prevailing environmental conditions across landscapes, making butterflies especially vulnerable to habitat fragmentation and/or extreme environmental disturbance (Bennett, 1999) in urban areas.Moreover, although butterfly communities are expected to be influenced by expanding environmental disturbances, which are most common in urban landscapes (Fenoglio et al., 2020;Kurylo et al., 2020;Mata et al., 2014;Tzortzakaki et al., 2019), slight changes in urban landscape management, such as improving habitat connectivity, have been suggested to increase butterfly richness, diversity and abundance in these systems (Diamond et al., 2023;Dylewski et al., 2019;Iserhard et al., 2019;Mata et al., 2014;Nagase et al., 2019).Thus, the increase in species richness, diversity and abundance may be more common among frugivorous butterflies in tropical systems that generally use multiple host plants as food sources (Nobre et al., 2012).Given their generality in plant food resources, frugivorous butterfly species make excellent models for understanding the local pattern of species' community responses to urbanisation, allowing us to test various hypotheses about the influence of environmental heterogeneity filters in tropical species' communities (Nobre et al., 2012).
Here, we sought to determine how (1) butterfly diversity varies with changes in environmental conditions in a city in Nigeria, (2) habitats with environmental heterogeneity in an urban gradients influence butterfly richness and abundance and (3) the local-scale butterfly diversity in an urban landscape varies across habitats with different microgeographic patterns.

Study area, butterfly collection and measurement of vegetative index
The field study was conducted in two urban (savannah woodland vs. gallery forest) habitats in and around the densely populated Federal University of Lafia, Nasarawa State, Nigeria (Figure 1).The city of Lafia is located at a latitude of 8 28 0 N and longitude of 8 32 0 E and 158 m above sea level, with temperature ranging between 26.8 and 27.0 C and an annual rainfall around 165 mm.Our field study location was selected because it is characterised by grassland savannah and scrub woodland interspersed by strips of gallery forests (Ombugadu et al., 2021).To understand local-scale biodiversity patterns in heterogeneous urban habitats of frugivorous butterflies in Lafia, we selected our two habitat types based on the defined vegetative structure between grassland savannah and gallery forests (Ombugadu et al., 2021).
Daily butterfly collection was carried out in a 6-week sampling period within the dry season (December 2021-January 2022), with both habitats visited every day.To avoid sampling bias and pseudoreplication, we used a systematic random points approach in the geographic positioning system (etrex GARMIN-12) to generate and mark butterfly sampling spots [created a 5 Â 5-m quadrant around each point] across habitats (Figure 1).For each week, we marked three sampling spots in each habitat (making a total of six points per week), with each spot separated by an approximate 100 m (depending on the availability of trees to suspend the traps) to ensure a representative coverage of the city.At the end of the study, we had 36 randomly selected spots, where butterfly traps were mounted in the two urban habitats, with 18 points each from habitat type (savannah woodland vs. gallery forest).In each of the 36 sampling spots, we only counted visited plants (by identifying plants with fruits and flowers to family level following Ombugadu et al., 2021) within our created sampling quadrants, which we used as a metric for host plant breadth in subsequent analyses.
In addition, we laced each trap with fermented palm wine and mashed rotten bananas (Freitas et al., 2014;Freitas et al., 2021;Hughes et al., 1998) to attract a large number of frugivorous butterflies across the two habitats.We mounted traps with fresh baits in the morning (07:00 AM) and inspected them in the evening (06:00 PM) daily.When collecting butterflies in the evening, traps were kept closed and abundances were recorded.For our collected butterflies, we had one voucher specimen collected for monomorphic species and two for sexually dimorphic species to enable us to use the recorded voucher collections necessary for onward taxonomical confirmation of species as identified in the field (Brattström, 2021(Brattström, , 2020;;Larsen, 2005aLarsen, , 2005b)).Aside from butterflies used as voucher specimens, all other butterflies trapped were released after being counted.The voucher specimens, kept in the Zoological Laboratory in Department of Zoology at the Federal University of Lafia, Nigeria, were preserved by injecting 75% ethanol into the abdomen to prevent infestations, allowed to air dry and were separated into the Savannah Woodland and the Gallery Forest species.

Collection of environmental data
We used the coordinate of the point for each spot butterflies were sampled during the field survey to collect the environmental data from the National Aeronautics and Space Administration (NASA) database (https://power.larc.nasa.gov).We obtained 1 Â 1 km scale environmental data (mean, minimum and maximum air temperature, wind speed and relative humidity) for our study area.We used the mean variability of these environmental variables across the array of 5 Â 5-m sampling quadrants.These environmental data were collected at 1-h intervals when sampling was done, and the average for each sampling site across the study period was determined for analyses to compute for the daily environmental variability (Ombugadu et al., 2021).We used "Envfit" function in VEGAN package to scale environmental variables into vectors that factored them onto ordinations to remove the effect of a possible relationship between climatic conditions across habitats (Oksanen, 2011).
We used these daily environmental variables as a proxy for measuring the local-scale responses of butterflies to the heterogeneous urban landscape in places with rapid marginal changes in climatic conditions (Pacifici et al., 2015).This is because these environmental variables can directly impact the dispersal ability, thermal physiology, development, phenology and reproduction of organisms, which can affect their diversity and abundance (Maclean & Early, 2023).

Statistical analysis
All data analyses were done in R Console software version 4.2.3 (R Development Core Team, 2023).To determine the diversity of butterfly communities in Lafia urban areas, we computed the gamma diversity index of butterfly communities across habitats using the diversity function in the VEGAN package (Oksanen, 2011).We estimated the Bray-Curtis measure of community diversity and visualised the dispersion of the community similarity structure using the resultant non-metric multidimensional scaling (NMDS) as implemented in the metaMDS function in the VEGAN package (Oksanen, 2011).We centred and scaled environmental data using the envfit function implemented in the VEGAN package (Oksanen, 2011) to avoid biassing the environmental variables due to differences in magnitude and scale between each environmental value across habitat types.To assess whether the prevailing environmental conditions across habitats significantly influenced species diversity, we used the scaled values of the environmental variables to conduct an analysis of similarities (ANOSIM) test using the anosim function in the VEGAN package (Oksanen, 2011), which we then visualised using an NMDS plot.We computed and estimated site area of habitats, which we used in subsequent analysis to account for the impact of habitat fragmentation on species diversity.
We estimated butterfly abundance in Lafia city as the total number of sampled individuals across all sampling spots in each habitat type (savannah grassland vs. gallery forest) by summing up the computed total diversity index of butterfly communities across habitats areas (the Bray-Curtis to estimate species dissimilarity).Next, we estimated butterfly total species richness as the cumulative number of butterfly species sampled from each habitat across habitats using the specnumber function with one margin in the package VEGAN (Oksanen, 2011).We constructed two generalised linear models with the Poisson family as implemented in the glmer.nbfunction implemented in the lme4 package (Bates et al., 2015) to assess the specific determinant factors influencing the local-scale biodiversity of butterflies across heterogeneous habitats in a city.For each model that we constructed, we used site identity as a random factor to account for the effect of a possible impact of sites, due to the possible overlap of sites that are within the 1 Â 1 km range (Figure 1) of environmental variables collected from the NASA database.By accounting for the effect of sites, we were able to directly evaluate the influence of habitat types and environmental conditions on the community of butterfly species when analysing for variance (Gelman, 2005).In the first model, we wanted to assess the influence of environmental conditions across heterogeneous urban habitats on local-scale species richness along an urban landscape gradient.Thus, we used the computed decomposed species (cumulative number of butterfly species sampled) richness index as a response variable with the week of collection, habitat type, weather condition during sampling (sunny vs. windy), maximum mean temperature of the dry season, minimum mean temperature of the dry season, vegetative indexes (plants with fruits and flowers) and daily humidity as our predictor variables while using the week of the collection as a covariate to assess the effects of weekly temperature variability on local-scale biodiversity.In the second model, we wanted to evaluate the impact of environmental conditions across heterogeneous urban habitats on local-scale species composition in the city.Thus, we fitted the second model with the Bray-Curtis measure of community diversity as the response variable with habitat type, weather condition during sampling (sunny vs. windy), maximum mean temperature of the dry season, minimum mean temperature of the dry season, vegetative indexes (plants with fruits and flowers) and daily humidity as our predictor variables while using the week of the collection as a covariate.For all our analyses, we conducted significance tests with a significance level of α = 0.05.

Effect of habitat patches on butterfly species diversity
We collected a total of 510 fruit-feeding butterflies from 20 species in the family Nymphalidae (Table S1).We found that butterfly species were highly significantly dissimilar between habitats (Figure 2).This resulted in the observed butterfly species diversity significant differences between urban savannah woodland and gallery forest habitat patches (Figure 3a), with butterfly species diversity higher in the savannah woodland when compared with the gallery forest (NMDS stress = 0.135).Although butterfly species diversity is positively associated with the number of plants across habitats (ANOSIM test: r = 0.078, p < 0.0001), this only explains 7.8% of butterfly community diversity.In addition, we found that butterfly community diversity negatively associated with large habitat patches with warmer temperature in the gallery habitat (Figure 3; Figure S1), indicating that small habitat patches with warmer temperature might influence higher species diversity in tropical urban habitats.

Effect of landscape heterogeneity on butterfly species community
We found evidence of variation in the butterfly community richness (χ 2 = 5.50, p = 0.019) and abundance (χ 2 = 5.96, p = 0.014) between the two habitats (savannah woodland vs. gallery forest), with higher butterfly species richness in the savannah woodland relative to the gallery forest (Figure 4a).This may have been associated with the warmer daily temperatures recorded in the savannah woodland through time (β = 1.964 ± 0.615, p = 0.001).In addition, there was evidence for the influence of habitat types (β = 0.415 ± 0.171, p = 0.015) on butterfly abundance, with greater numbers found in the savannah woodland than in the gallery forest (Figure 4b).There was, however, no evidence of prevailing environmental conditions influencing butterfly diversity (Figure S2) or abundance in the city (Table 1).Furthermore, we did not find any relationship between daily humidity recorded and butterfly species richness in the savannah woodland (Table 1).
F I G U R E 3 Results from the non-metric multidimensional scaling (NMDS) of (a) dispersion of butterfly community composition and structure across two habitats in a tropical urban landscape, (b) daily temperature variation influencing butterfly species community composition across habitats and (c) butterfly species diversity and habitat patch size relationship across a tropical urban landscape.These plots have 510 fruit-feeding butterflies collected from 36 collection spots across two urban habitats (savannah woodland vs. gallery forest).Coloured points represent butterflies collected, whereas grey trend lines represent the community structural relationship and environmental factors of the geographical space occupied by the butterflies.Maximum_temp, Maximum temperature.
F I G U R E 2 Results of the non-metric multidimensional scaling analysis on species dissimilarity across two urban (savannah woodland vs. gallery forest) habitats within the Lafia City, North-Central Nigeria.

DISCUSSION
We found evidence that environmental heterogeneity between habitat patches (savannah woodland vs. gallery forest) in tropical urban areas influences the composition and structure of butterfly community composition.Specifically, we found that small habitat patches (savannah woodland) had higher richness and abundance of butterfly species when compared with large habitat patches (gallery forest).We found evidence that the number of host plants (number of plants visited by butterflies) marginally impacted butterfly species diversity, but not richness or abundance.However, we found that warmer temperature associated with small habitat patches (e.g., savannah woodland) supported the high richness and diversity.Thus, our study demonstrates that the effect of urban landscape heterogeneity, measured as the habitat types across an urban area, is influencing butterfly community composition and structure, with small habitat patches, such as savannah woodlands, potentially serving as a habitat refuge (i.e., habitat attracting butterfly species) in tropical urban areas.

Impact of urban fragmented habitat heterogeneity on frugivorous butterfly community structure
Previous studies have suggested that urbanisation, which tends to create fragmented habitats, negatively impacts the capacity for species to persist in urban environments (Rands et al., 2010), thereby reducing biodiversity (Haddad et al., 2015).However, recent studies F I G U R E 4 Change in richness and abundance of fruit-feeding butterfly communities across two habitats in an urban landscape, with (a) savannah woodland having higher fruit-feeding butterfly community richness than the gallery forest, and (b) savannah woodland having higher fruit-feeding butterfly community abundance than the gallery forest.Red-coloured scattered dots represent raw species richness and abundance with 510 fruit-feeding butterflies collected from 36 collection spots across two habitats, whereas predicted values ±1 standard error of estimates from the generalised linear regression are represented by connected dots with error bars showing the standard error of the means.
T A B L E 1 Statistical parameters from generalised linear mixedeffects models of butterfly species community composition with richness index and abundance index in urban habitats in Lafia.have shown that certain urban habitats might support high insect species diversity and richness (Diamond et al., 2023).In line with this recent evidence, we found that habitat types in tropical urban areas are differently influencing the community composition and structure of fruit-feeding butterflies, with savannah woodland habitats supporting higher diversity and richness when compared with the gallery forest habitats within the same urban setting.Thus, our findings offer support for the debate that some urban habitats are able to harbour diverse assemblages of insects (Diamond et al., 2023;Fontanarrosa et al., 2009).In addition, the results of our study further indicate that local-scale biodiversity may be higher in some areas relative to others within a heterogeneous urban landscape or human-impacted habitats (Vogel et al., 2023).For example, studies have indicated that urban habitats offer refuges for the establishment of insect populations, such as lawns excluding non-native vegetation (Smith et al., 2015), gardens (Gunnarsson & Federsel, 2014), habitats impacted by agro-ecological processes (Vogel et al., 2023) and parklands (Sing et al., 2016).This is because not all areas in human-impacted landscapes are created with equal support for insect community structure.Of course, a variety of factors related to urban development are likely involved in shaping differences in diversity and richness of insects across habitats.In the case of frugivorous butterflies, it may be that the savannah woodlands, with relatively higher richness and diversity of insects, have more suitable environmental conditions (Kati et al., 2012;Molina-Martínez et al., 2013).There are a variety of factors which may allow for a greater diversity of insects in some urban habitats over others, such as plant diversity.Maintaining urban habitats with a diverse assemblage of native vegetation types, rather than commonly non-native ornamentals, may promote maximum insect biodiversity (Burghardt et al., 2009).Thus, future studies should involve identifying the specific environmental mechanisms that shape insect species richness and diversity across heterogeneous urban landscapes such that habitats can be maintained that both meet human needs and allow for diverse insect assemblages.
Although the majority of habitat parameters tested here were not associated with butterfly community structure, we found that daily maximum temperature variation among urban habitat patches might be a driving mechanism behind the differences in butterfly richness and diversity.This finding is consistent with previous studies that showed how hotter microclimates in the urban landscape are associated with higher species richness of insect taxa (Adams et al., 2020;Uhler et al., 2021).Perhaps, this pattern is potentially due to the general association between warmer climates with high dispersal rates due to thermal range flexibility (Chown & Duffy, 2015), phenology (Altermatt, 2012) and higher biodiversity (Allen et al., 2002).Thus, we speculate that the savannah woodland habitat likely offers a warmer overall microclimate in the urban landscape than the gallery forest, promoting higher relative species richness of butterflies.However, although the positive association between temperature and biodiversity was supported for these frugivorous butterflies, not all taxa appear to follow this trend.For urban ant communities (Pelini et al., 2014), richness has been found to decline with rising temperatures, and for phorid flies (McGlynn et al., 2019), richness peaks at intermediate temperatures and evenness declines in warmer microclimates.Furthermore, responses of insect community structure across temperatures in urban habitats are also likely mediated by the interaction between increasing temperature and latitude (Diamond et al., 2023).Thus, as climate change intensifies, we may expect the community composition and structure of some insect groups in urban landscapes to shift (Stefanescu et al., 2004), with variations in responses across latitudes (Okoro et al., 2023;Pelini et al., 2014;Perez et al., 2022).

Impact of landscape heterogeneity on frugivorous butterfly community composition
We found significant difference in the abundance of butterflies across two habitats with heterogeneous microgeographic landscapes in an urban area, with higher butterfly abundance found in the savannah woodland than the gallery forest.This finding is in line with assessments of community composition for other insects in urban areas of Africa, such as mosquitos varying in abundance across different habitat patches in an urban-rural gradient (Johnson et al., 2020;Okoro et al., 2023).However, unlike the species richness index, the difference in the average abundance of butterflies in our study was not associated with any recorded environmental factors across habitats.This is surprising because previous studies have indicated that arthropod communities vary in composition based on factors assessed in this study, such as humidity (Hamřík et al., 2023;Pelini et al., 2014) or density of vegetation (Neves et al., 2023;Pendleton et al., 2011), providing evidence that different mechanisms in heterogeneous landscapes shape different arthropod groups.Perhaps, the difference in abundance of these fruit-feeding butterflies may be associated with other habitat characteristics, such as canopy coverage (Hendy et al., 2020;McCabe et al., 2019), which likely differs throughout heterogeneous urban landscapes.
Despite our lack of knowledge regarding which environmental mechanisms are shaping frugivorous butterfly community composition in this urban landscape, our results indicate that the heterogeneity introduced by urban expansion is related to differences in relative abundance of these insects across sites.Evidence already indicates that distance from urban centres is positively associated with the abundance of flying insects such as butterflies (Callaghan et al., 2021), potentially due to more impervious surfaces being found near urban hubs that impede plant growth (Lagucki et al., 2017;Piano et al., 2020).However, our finding on the influence of habitat types on butterfly abundance indicates that urbanisation is probably converting some habitats of insect communities that previously maintained homogeneous abundance throughout tropical urban landscapes into patches of modified habitats, among which abundances vary.For example, in an urbanised tropical forest system in Brazil in which habitat fragments vary in their vegetative characteristics, relative abundances of different dung beetle species were found to vary across the heterogeneous landscape (Korasaki et al., 2013).In addition, within these urban habitat fragments, some species are experiencing rises in abundance, whereas others are declining with increasing urbanisation (Rickman & Connor, 2003).Therefore, conservation efforts for insects in urban environments should focus on maintaining high abundances of species of concern across the heterogeneous landscape, which will likely involve taking different approaches based on habitat type and balancing the needs of different species.

CONCLUSION
In summary, we showed that in a community of butterflies in a tropical urban landscape found in Nigeria, diversity, richness and abundance varied between sites in a heterogeneous urban landscape.
Specifically, measured butterfly community indexes (diversity, richness and abundance) were higher in small patches of savannah woodland habitat relative to the gallery forest habitat, with the local-scale biodiversity of these insects reduced in the large gallery forest habitats.Thus, our study indicates that butterfly communities are able to persist in some fragmented urban habitats.Although only an effect of warmer temperature was evidenced for increased butterfly diversity and richness in this study, future studies should involve investigations of other environmental mechanisms that work to shape differences in insect community structure and composition for the sake of improving conservation of urban insect populations.
One question of specific interest will be to explore how pollination rates in tropical urban areas will determine the community composition and structure of these butterflies and other insects within heterogeneous habitat patches mostly created by increasing human activities in urban areas.Because our findings clearly showed the influence of daily moderate temperature and small habitat patches across urban spaces on the distribution of tropical butterflies.

F
I G U R E 1 Study area of the Lafia city, North-Central Nigeria, denoting the sampling spots.All sampling spots in Lafia city with different microhabitats are shaded in different colours.Labelled sampling spots showing 36 areas of 5 Â 5-m quadrant with 100-m buffers between sampling points where traps were set.
Thus, there would be a need for a long-term, extensive and systematic insect monitoring programme for butterflies across fragmented habitats that are either disturbed (open, natural reserves areas or unprotected areas) or undisturbed (protected) and are possibly harbouring diverse insect species for evidence-based conservation managements of tropical butterfly biodiversity.