Species List and Temporal Trends of a Butterfly Community in an Urban Remnant in the Atlantic Forest

Abstract

The Brazilian Atlantic Forest is currently reduced to a few, small fragments surrounded by anthropic landscapes. Urban forest remnants play an important role in housing biodiversity in urbanized areas and enabling species dispersion between larger natural areas. Describing and monitoring the biodiversity in these anthropized environments is a useful tool for urban ecology and species conservation. By compiling long-term data, this study proposes a species list of tropical diurnal butterflies for an Atlantic Forest remnant in Southeastern Brazil and describes the temporal diversity patterns of the community. Inventories were carried out from the 1970s to 2000 to record butterfly incidence data in a 250 ha fragment of seasonal, semideciduous Atlantic Forest, resulting in a species list of 706 species presented here for the first time for the studied area. From 1998 to 2000, weekly standardized observations enabled inferences on temporal trends in butterfly diversity. Total species richness peaked in the climatic transitions, and a cycle of 52 weeks was reported for beta diversity. Butterfly families lack synchronized temporal fluctuations. Nymphalidae and specifically the fruit-feeding nymphalids were considered good surrogates for short-term studies and monitoring community changes.

1. Introduction

The Brazilian Atlantic Forest is a global biodiversity hotspot that has been replaced by small, isolated fragments that nowadays sum up to 12% of its original occupancy area due to the expansion of urban areas and agricultural frontiers [1,2,3]. The overexploitation of this forest led to a current patchy landscape, forest remnants susceptible to intense edge effects [3,4,5], fire, species extinction [6], changes in biodiversity distribution patterns [7,8] and ecosystem function loss [9,10]. Forest fragments are often too isolated from each other and surrounded by inhospitable anthropic landscapes, impeding species dispersion and not supporting viable populations in the long term [11]. Forest remnants in urban areas play a crucial role in supporting biodiversity and facilitating species movement between larger natural areas [12]. While some species can adapt to urban environments, well-connected urban green spaces are essential for those species that face challenges due to urbanization and need to migrate to more suitable areas [13,14]. Therefore, describing and monitoring biodiversity in these human-modified environments is important to understanding which species can thrive in or near urbanized areas. This knowledge can help us make cities less hostile to biodiversity and mitigate species loss due to urban expansion [15].

The destruction of the Brazilian Atlantic Forest and its transformation into a fragmented landscape has been ongoing since the arrival of European colonizers [3]. However, there is limited information on the effects of this destruction on insect populations and their resilience to the altered environment. Long-term monitoring programs that focus specifically on insects have only recently been established in Brazil, providing reliable data that spans no more than 20 years regarding changes in abundance and diversity [16]. In this study, we aim to present valuable historical, unique, and systematic data with a high level of taxonomic resolution. This information is intended to support current and future monitoring efforts, as well as short-term studies in the Atlantic Forest.

Butterflies stand out as indicators for inventorying and monitoring biodiversity and forest integrity in fragmented landscapes [7], especially due to their sensitivity to microclimatic factors, short life cycle, and ease in sampling and identification in the field [17,18,19]. In the present study, diurnal butterflies were recorded from the 1970s to 2000, with the aim to (i) obtain a species list for a 250 ha urban Atlantic Forest fragment, (ii) describe their temporal ecology to identify optimal temporal windows for biodiversity sampling in the seasonal Atlantic Forest, and (iii) select butterfly groups as surrogates for biodiversity conservation initiatives based on which group better represents temporal changes in the overall community. Although these insects are well established as biological indicators, their pronounced diversity in the tropics makes it necessary to focus on specific taxa to develop viable and efficient conservation strategies. The whole community and specific butterfly groups are expected to respond to climatic seasonality, with increasing species number in the warm-rainy season and species composition changing between seasons. Thus, we can identify adequate periods for short-term diversity studies and long-term monitoring in the seasonal Atlantic Forest.

2. Methods

2.1. Study Area and Sampling

This study was carried out in “Área de Relevante Interesse Ecológico Mata de Santa Genebra” (hereafter MSG) (22°49′ S, 47°07′ W), a 250 ha fragment of the Atlantic Forest in Campinas municipality, São Paulo State, Brazil, located in an agricultural matrix [13,20,21] (Figure 1). Despite being over two decades since the data were collected, all forest fragments are still present. Additionally, aside from a small expansion of urban areas, secondary forests have expanded in regions of old abandoned pastures (AVLF pers. obs). The local climate shows two marked seasons: a warm-rainy season from October to March and a cold-dry season between April and September [22]. From 1998 and 2000, the average temperature was 29.6 °C for the warm-rainy season and 26.9 °C during the cold-dry season, and mean precipitation was 202.6 mm and 46.2 mm, respectively (Figure 2), which is considered typical for the area [22]. The vegetation is classified as semideciduous seasonal forest, in which some of the trees lose their leaves in the cold season. The area is mostly level and uniform (semideciduous secondary forest with some swampy headwaters). The MSG is inserted in a complex landscape composed of urban and agricultural lands, increasingly impacted by suburban development on its eastern side, and connected in different degrees to small forest fragments from less than one to nearly 200 ha (see details in Brown and Freitas [13] and Figure 1).

2.2. Butterfly Sampling

The butterfly surveys conducted at MSG by KSB (since the 1970s) and AVLF (from 1988 to the present) have allowed us to propose here, for the first time, a species list of six diurnal butterfly families for the area (the family Hedylidae was not included since they are nocturnal and better sampled with light traps). We follow Heikkilä et al. [23] for Papilionoidea, and the taxonomy of butterflies follows mostly Lamas [24], updated after Wahlberg et al. [25] for higher classification of Nymphalidae, Tyler et al. [26] for Papilionidae, the Riodinidae Species Checklist [27] and Zhang et al. [28] for Riodinidae, Robbins et al. [29] and subsequent publications for Lycaenidae, Zhang et al. [30] and Leong et al. [31] for updates in Pieridae, the Euptychiina Species Checklist [32] for Euptychiina, Paluch [33] for Actinote, and Nuñez et al. [34] and Penz [35] for Agraulis and Dryas. We follow the higher classification of Hesperiidae summarized by [36], and the taxonomy of the group follows Zhang et al. [37] with recent updates [38,39,40,41,42,43,44,45,46,47].

Between January 1998 and December 2000, standardized maximized daily censuses were conducted following Brown [48]. These censuses were carried out in different environments and trails within the area (fragment edge, interior and marsh), with two experienced collectors using entomological nets (with a handle of 3 m in length) and attractive traps, from 7:30 a.m. to 6:30 p.m. on sunny days (from 8:30 a.m. to 5:30 p.m. during short days in winter months), with observation times varying between 1 and 12 h depending on weather conditions. In each census, the species and observation time were recorded. On sunny days, each hour was considered a full hour, and during cloudy periods or days, each hour was recorded in half since butterfly activity decreases in these conditions; rainy periods were not considered. The daily censuses (hereafter partial lists) of the same week were combined into lists of 14 h of observation (hereafter complete lists), each one being considered in this study as a sampling unit, summing up 89 lists in total (more details in Iserhard et al. [49]). Only individuals that were difficult to identify were collected, taken to the laboratory for correct identification, and later incorporated into the Zoological Collection of the Museu de Diversidade Biológica of the Institute of Biology of the State University of Campinas (ZUEC).

2.3. Temporal Analysis

Sampling units (complete lists, 89 in total, see above) were compared by species richness, frequency, and beta diversity across weeks, months, and seasons (dry and rainy). Species prevalence was obtained by calculating the number of complete lists in which each species was recorded as follows: (1) “common”—present in 40 complete lists (45%) or more; (2) “uncommon”—present from 18 (20%) to 39 complete lists; (3) “erratic”—present from 6 (5%) to 17 complete lists; and (4) “visitors”—present in 5 or less complete lists.

Species richness in the three sampling years (1998, 1999, and 2000) was plotted against the sample-based rarefaction curve for the total sample, which gave the expected species richness in a random subset of any particular size, following DeVries and Walla [50]. The statistical significance of these comparisons was evaluated using 95% confidence intervals for the total rarefaction curve. Species richness for total community was estimated using Chao2 and Jackknife1 indexes, indicated for incidence data [51]. The number of monthly visits in the three years combined was not balanced, varying between 4 and 14. Therefore, to understand in which month a significantly higher number of species was recorded, rarefaction and extrapolation curves by samples were obtained for each month, with a maximum number of samples of 14 weeks. In this way, the species number could be compared in a scenario of an even number of visits. The statistical significance of all rarefaction curves was evaluated using 95% confidence intervals. Rarefaction curves and estimators were obtained using EstimateS 9.1 software [52].

We calculated the species number in each partial list and combined the sums of species richness from partial lists in the same week, obtaining 859 records of butterfly species observed during 1 to 14 h. To test whether there is a relationship between hours of observation (predictor variable) and the observed species richness (response variable), regression analyses for total community and each butterfly family were performed in R 3.5.1 software [53]. To understand in which month the relation species/observation time was maximized for the entire butterfly community and each family, we plotted the number of hours of observation by the species richness observed, numbering each point with the corresponding month of observation, and analyzing it graphically.

To test community change over time (beta diversity), we calculated Jaccard indexes for different combinations of pairs of the 89 sample units and plotted the mean pairwise similarity values against the time intervals between paired lists, following Brown [54]. High mean values indicate high similarity among samples, being interpreted as low species replacement. These analyses were developed using the PAST 3.04 software.

To identify which groups showed less variation in their proportions of observed richness compared to the total community richness, and could therefore be considered targets for conservation studies, the proportion of each group in relation to the total community for each list was calculated, and then the coefficient of variation (CV) of the proportions in the 89 lists was determined. The coefficient of variation is the ratio between the standard deviation and the mean. A low coefficient of variation represents similar proportions among samples, providing a stable estimate of the number of butterfly species in the area in any sampling event. Each family was analyzed along with two groups of Nymphalidae that are frequently used in conservation studies: the tribe Ithomiini [17,55,56] and the subfamilies of fruit-feeding butterflies (reviewed by Freitas et al. [57]).

3. Results

This study presents the first butterfly species list for Mata de Santa Genebra (Appendix A), compiling 50 years of observations (from the 1970s to the present). A total of 706 butterfly taxa were recorded from the following families: Hesperiidae (314 species, 44.5% of the total), Nymphalidae (215 spp, 30.5%), Lycaenidae (89 spp, 12.6%), Riodinidae (43 spp, 6.1%), Pieridae (28 spp, 3.9%), and Papilionidae (17 spp, 2.4%). This sampling effort tended towards an asymptotic rarefaction curve (Figure 3), and it has likely captured a representative portion of the butterfly diversity present in the community.

Based on species prevalence from samples between 1998 and 2000, the entire community presented 35.6% of common species, 17.8% uncommon, 18.4% erratic, and 28.2% visitors (

Table 1. Species prevalence for all butterflies and each family. For the definition of categories, see the Methods section.

	Common	Uncommon	Erratic	Visitors	Total
All butterflies	228 (35.6%)	114 (17.8%)	118 (18.4%)	181 (28.2%)	641
Hesperiidae	88 (30.7%)	55 (19.2%)	59 (20.6%)	85 (29.6%)	287
Nymphalidae	89 (45.9%)	28 (14.4%)	28 (14.4%)	49 (25.3%)	194
Lycaenidae	12 (15.6%)	18 (23.4%)	18 (23.4%)	29 (37.7%)	77
Riodinidae	10 (25.6%)	9 (23.1%)	6 (15.4%)	14 (35.9%)	39
Pieridae	20 (71.4%)	2 (7.1%)	5 (17.9%)	1 (3.6%)	28
Papilionidae	9 (56.3%)	2 (12.5%)	2 (12.5%)	3 (18.8%)	16

). Combining the two first categories as “frequent” and the two last as “rare”, three patterns could be observed for the butterfly families: (i) balanced number of frequent and rare species (Hesperiidae and Riodinidae); (ii) more frequent species (Nymphalidae, Pieridae and Papilionidae); and (iii) more rare species (Lycaenidae) (Table 1).

Comparing the number of species in each year, 1998 was slightly less species-rich, while 1999 and 2000 did not differ from the total richness for MSG (Figure 3). The Chao2 and Jackknife1 estimated maximum species richness were 689 (mean = 677.2, SD = 12.21) and 710 (mean = 701.3, SD = 8.85) species, respectively, numbers very close to the actual species richness of 706 species in the study area.

Temporal Patterns

For the temporal analyses, only 641 butterfly species (90.7% of the total richness) recorded between January 1998 and December 2000 were considered, totaling 287 Hesperiidae, 194 Nymphalidae, 77 Lycaenidae, 39 Riodinidae, 28 Pieridae, and 16 Papilionidae species. In 1998, a biannual pattern of species richness was reported for the total community, while a single peak occurred in April in both 1999 and 2000 (Figure 4A). These patterns were particularly influenced by the two most species-rich families, Hesperiidae and Nymphalidae (Figure 4B). In general, there was a remarkable peak of species richness at the end of the rainy season (March-April), followed by a decrease in species throughout the dry season, with occasional, stochastic increases in species observation during this period. Three distinct patterns emerged from temporal variation on the family level: (i) in Hesperiidae, Nymphalidae and Riodinidae, species richness peaked in climatic transitions (type I—seasonal, according to [58,59], Figure 4B); (ii) two families presented higher richness throughout one whole season: Lycaenidae during the dry season and Papilionidae during the rainy season (also type I—seasonal, Figure 4C); and (iii) the Pieridae showed no marked differences in richness between seasons (type III—aseasonal, Figure 4D).

Species richness was positively correlated to hours of observation for all taxa (F_(2,856) = 422.0, p < 0.0001, R² = 0.50), this relation being equally strong in Nymphalidae (R² = 0.52) and very weak in Papilionidae (R² < 0.20) (Figure 5). The number of species recorded in a complete list varied from a minimum of 153 (12 October 1999) to a maximum of 334 (7 April 1999), and April stood out as the month with the higher richness recorded in partial and complete lists for the total community (Figure 5 and Figure 6), as well for Nymphalidae, Hesperiidae and Riodinidae. Lycaenids were mostly found in July partial lists, Papilionidae in rainy season months, and Pieridae did not show a clear pattern (Figure 5).

By extrapolating monthly taxonomic diversity across 14 samples and comparing balanced sample efforts for each month, we found that the estimated species richness peaked in March and April (Figure 6). This observation supports the idea that the transition between the rainy and dry seasons is particularly rich in species within the MSG. Although the confidence intervals overlapped for most months, indicating that butterfly species richness varied slightly over time, more species were recorded during the mid-dry season than in the mid-rainy season (Figure 6).

Butterfly community composition changed smoothly throughout weeks in a sinusoidal pattern, reaching minimum similarity between samples 26 weeks apart (independent of the year and month), and maximum similarity every 52 weeks (sampling at the same time of the year in distinct years) (Figure 7). All families showed similar temporal beta diversity patterns, except Pieridae, for which similarity means were the highest and most homogenous throughout the weeks, showing a low temporal beta diversity. Weekly samples of Lycaenidae and Riodinidae were never more than 50% similar, suggesting a less predictable species composition on a time scale, with different species being recorded in the same month in different years.

Nymphalidae and Hesperiidae species represented close to 40% of the community species each, with this proportion changing by 10% and 13% over the weeks, respectively. Fruit-feeding nymphalids represented 17% of the total diversity, varying by less than 20% in all samples. All other families and Ithomiini counted for less than 10% of total species richness, this proportion changing by at least 27% compared with total butterfly richness, except in Pieridae, whose representativeness varied as much as fruit-feeding species (Figure 8).

4. Discussion

4.1. Butterfly Diversity at MSG and Community Dynamics

Based on all available lists, the total butterfly species richness in the MSG is 706 species, which is a higher count than that reported in areas up to 20 times larger within the Atlantic Forest [60]. Although several species are transitory and some species documented in the 1970s may now be locally extinct, the overall species richness remains impressively high. Over three years of intensive sampling conducted for this study (from 1998 to 2000), a total of 641 species were reported (90.8% of the actual species richness), a figure that is still comparable to the diversity found in several large, continuous areas of the Atlantic Forest [54,60,61] and unpublished data.

This high local species richness in MSG is sustained by a complex dynamic of metacommunities, characterized by significant exchanges of species and individuals present among nearby fragments [13,17]. In the present study, only 35.6% of the species were classified as “common”, and some species were only present during certain seasons and years, especially in the families Hesperiidae and Lycaenidae. Furthermore, many species in all butterfly families are transient, meaning they appear and disappear from the area over periods ranging from days to decades [17] (pp. 937–939). Brown [54] presented a table displaying the occurrence patterns of 37 butterfly species in the MSG categorized as “erratic” (not rare residents or accidental visitors, as in the present study). Several of these species are seasonal migrants and maintain stable populations in other nearby fragments in the region [13,17]. For instance, the “erratic” metalmark butterfly Catocyclotis malca (Schaus, 1902) (Riodinidae) was reported in only 11 out of the 89 complete lists in MSG, while a resident population of this species exists in a nearby swamp habitat just 30 m away from the southern limits of the study area (AVLF, pers. obs.). Some local extinctions have also been reported over the five decades of surveys. Two notable examples are the butterflies Pierella nereis and Opoptera syme (Nymphalidae: Satyrinae). These two species were reported in the first inventories in the early 1970s, but have not been sighted again in the MSG (including the three years of intensive sampling in the present study), suggesting they may be locally extinct. However, both species are still present in a similar-sized fragment located 16 km to the west of MSG (Ribeirão Cachoeira, 207 ha, “RC” in Figure 1; see details in Brown and Freitas [17]).

A butterfly inventory in several areas in the eastern Atlantic Forest showed that new species are added to the list each additional day of sampling, while other species seem to disappear ([62], see also Brown [54] (Table 10.6)). This suggests that the dynamics of species replacement reported here are not exclusive of MSG, but likely a general pattern for butterfly communities in the Atlantic Forest (as discussed by Brown and Freitas [17]).

In the context of biodiversity monitoring, understanding the results obtained is crucial. While the frequent establishment of butterfly populations can lead to greater randomness in the local community, this effect is somewhat balanced out by the larger number and variety of species present. Each species indicates its specific resources and habitats, contributing to the overall understanding of the ecosystem. The butterfly communities in the Atlantic Forest appear to be open and fluid, influenced more by ecological factors than by evolutionary constraints [63,64]. Therefore, the presence of a particular butterfly species in a given locality may serve as a better indicator of connectivity and ecological conditions, rather than a reflection of their historical origin or long-term residence in that area (see also Brown and Freitas [17]). The present study confirms this pattern in a small forest fragment.

The current results highlight the significance of small urban and suburban areas in supporting regional biodiversity. Here, the MSG not only acts as a stepping stone for butterfly movement in the landscape, but also serves as a refuge, helping to preserve populations of hundreds of butterfly species.

4.2. Temporal Patterns

In general, the butterfly community was seasonal, with a higher number of species recorded in climatic transitions, particularly rainy-to-dry season. This biannual pattern for butterfly species richness has been widely reported in the Southern Atlantic Forest (from the states of São Paulo and central Minas Gerais to Rio Grande do Sul) [62,65,66,67]. Various biotic and abiotic factors, such as the increase in natural enemies, host plant senescence due to dry conditions, and temperature drops, can lead to lower species richness during the dry-cold season [61]. Additionally, community composition changes seasonally, with species similarity ranging from about 45% to 65% depending on the time interval considered (Figure 7). This noteworthy temporal beta diversity should be considered in biodiversity studies within the Southern Atlantic Forest.

Long-term temporal studies of Neotropics butterflies are limited, especially those focusing on all families (see Freitas et al. [68]). Ebert [62] followed butterfly communities for nearly a decade in three areas of the Atlantic Forest, Pozo et al. [69] carried out three years of sampling in a Mexican lowland tropical forest, and Brown [61] regularly visited a seasonal forest in the Southern Atlantic Forest (100 km apart from MSG) over nearly a decade. All these studies reported the lack of synchronized temporal fluctuations among butterfly families.

The non-seasonal variation in Pieridae species richness found in the present study is similar to what was reported by Pozo et al. [69] in Mexico. Additionally, the biannual pattern of species richness reported for Hesperiidae and Nymphalidae is similar in both studies. However, in Mexico, these families were linked to specific seasons, in contrast to the association with climatic transitions observed in the current research. While Papilionidae was associated with the dry-rainy season in Mexico, this family was found throughout the rainy season at MSG. This Atlantic Forest pattern may be due to increasing temperatures and rainfall that disrupt winter pupal diapause, as suggested by Brown [61]. The fluctuations observed in Lycaenidae and Riodinidae cannot be directly compared because these two families are combined in Pozo et al. [69]. Consequently, the specific temporal patterns of lycaenids could be obscured by a potential biannual variation in Riodinidae, as reported in the current study. Describing the temporal patterns of different butterfly families opens a window to understanding the relationship of these species with the temporal dimension of the environment, allowing for better predictions of their responses to climate change. Our data encourages exploring temporal ecology in the tropics to understand biodiversity life strategies in seasonal tropical environments.

The current findings align with the idea that tropical insect species adjust their life cycles to seasonal weather patterns [70]. Typically, tropical species synchronize their adult activity to warm seasons, when food resources are abundant and abiotic conditions are optimal (review in Freitas et al. [68] for the Neotropics). The community temporal patterns can be determined by the richer families following the optimal environmental conditions (Nymphalidae, Hesperiidae, Riodinidae, and Papilionidae in MSG), although some groups can display different temporal patterns by developing strategies to thrive during less suitable seasons (Pieridae and Lycaenidae in MSG, see also Kishimoto-Yamada and Itioka [70]). In the Southern Atlantic Forest, warm waves during the cold-dry season (“veranicos”) can enable butterfly activity and availability of food resources, for example, lycaenids exploring floral resources during the dry season in the MSG [17,61].

Neotropical butterfly communities are very dynamic, and butterfly species are reported to appear and disappear from MSG at irregular intervals, including purportedly sedentary species (see above). This explains the high beta diversity found for most butterfly groups throughout the three years analyzed in the present study, especially Hesperiidae, Lycaenidae, and Riodinidae (Figure 7). This decreasing similarity was also recorded over seven years in the MSG butterfly community [49]. On the other hand, a 13-year fruit-feeding butterfly monitoring in a larger fragment 100 km apart from MSG has shown predictable numbers of butterfly species richness throughout months and seasons, even after macroclimatic oscillations (El Niño-La Niña [68]). It remains to be studied if the temporal dynamic of Neotropical species is more intense in small forest fragments, where ecological pressures on biodiversity can be more intense due to higher susceptibility to disturbances (e.g., vegetation structure alterations) that strongly shape community and richness [54].

4.3. Implications for Inventories and Monitoring

The distinct temporal patterns observed in butterfly families underscore the importance of sampling over an entire year to accurately characterize butterfly biodiversity. However, many biodiversity assessment protocols are constrained by financial or staffing limitations, resulting in short-term studies that fail to account for seasonal variations or individual population changes. This often leads to biased reports that overlook the temporal effects of population dynamics [71]. To reduce these biases in short-term studies in the Southern Atlantic Forest, our findings suggest incorporating both climatic transitions into sampling designs, as this approach is expected to reveal up to 45% of known butterfly species.

In the present study, April was the month with the highest species lists in partial and complete lists (Figure 5 and Figure 6), with lists above 300 species obtained in some years and a maximum of 334 species in one complete list (see the Results section). These figures are not far from the maximums possible to be obtained in the tropics [48] and are above similar daily lists obtained in other neotropical sites [48,72]; A.V.L.F. unpublished. For example, the study by Iserhard et al. [49] in MSG considered only April lists, from 1997 to 2003, and the results were very congruent with the present, temporally detailed study. Brown and Freitas [13] pinpointed the blossoming in the region as a driver to congregate many butterfly species in MSG, showing that adult resources can be determinant to colonization and establishment of species in forest fragments. Therefore, butterfly observation in the region can be optimized in April when scheduling short-term studies, monitoring programs, and citizen science events applied to conservation and public awareness [73]. However, this approach does not sample the univoltine butterflies that are active only during the summer months (late November to early March), and therefore is not recommended for obtaining accurate local inventories.

4.4. Surrogates to Biodiversity Conservation

Butterflies are generally considered good biological indicators [18,57], but as shown here, not all families represent the temporal pattern of the entire community. In this sense, in the MSG, Hesperiidae, Pieridae, Nymphalidae, and fruit-feeding butterflies showed greater potential to be used as biological indicators since they present greater predictability over time and optimization of field observation, minimizing efforts for an accurate description of local biodiversity.

Two groups of butterflies, despite presenting a stable proportion of the community over the weeks, were considered unsuitable for this purpose: (1) Hesperiidae, because they are difficult to identify without the help of experts and are subject to low species detection even with intense sampling effort (Figure 5); and (2) Pieridae, because its temporal pattern did not represent total community fluctuation. Therefore, the use of these two families as indicators is not recommended.

Nymphalidae, in turn, showed the greatest temporal detectability (here accessed through the percentage of richness observed over the weeks). Even with the marked seasonal pattern of distribution and beta diversity, nymphalids represented on average 38% of the species in the MSG community throughout all weekly samplings, higher than the proportion of 29% estimated by [13]. These results, associated with a relatively stable taxonomy and ease of identification [13], support the use of this family as a bioindicator. Furthermore, in April, suggested here as the best time to conduct rapid studies, nymphalids were easily observed (see also [49]). For a robust sampling of this family, it is essential to combine sampling methods such as entomological nets and attractive traps that allow access to a greater number of species. In this sense, selecting only fruit-feeding nymphalids, also listed here as good indicators of the temporal variation of the total butterfly community, is interesting in projects with limited financial and human resources, since this group presents methodological advantages of standardization and comparability [57,69]. The smaller number of species in relation to the entire Nymphalidae and distinct morphological patterns at the subfamily level allow for monitoring and rough but informative identifications to be conducted by non-specialists [74], with the help of field guides.

Finally, Beccaloni and Gaston [55] suggested that the tribe Ithomiini (Nymphalidae: Danainae) represents a relatively stable proportion of 4.6% of local butterfly richness, reinforcing its potential as biodiversity surrogates in conservation initiatives. However, subsequent studies showed that this is not the case, as the proportion of ithomiines can vary from 2% to 8%, depending on the region [17,54], and the coefficient of variation of this proportion throughout time can reach almost 30% for this group (present study).

5. Conclusions

This study provides a comprehensive list of the entire diurnal butterfly community of MSG, enhancing our understanding of Lepidoptera biogeography, macroecology, and urban ecology. Furthermore, the historical and systematic data presented here can be considered as a validation baseline to recent and future long- and short-term studies, allowing for a better understanding of changes in biodiversity due to current climatic and urban conditions in the tropics. Although similar recent data is not available for the study area, the patterns described here align with findings from other studies in the region (see above) and increase our understanding of the dynamics of tropical butterfly assemblages, in a way that we hope to inspire new research and initiatives that will further advance the knowledge of temporal ecology. Additionally, we propose guidelines aimed at enhancing the quality of short inventories and public outreach activities, ensuring that our impact extends beyond science to include political advocacy and biodiversity awareness initiatives.