We monitored physical and chemical variables of both water and sediments in six low-order streams and two shallow lakes in the AGCV for seven years during the dry (August–September), rainy (February–March), and transition seasons (May and November; Fonseca et al., 2014). We used a Surber sampler placed on the streambed to collect macroinvertebrates by washing and rubbing the substrate through the sampler and a plankton net to sample microinvertebrates (mesh size of 80 μm). We also performed manipulative experiments to understand the influence of nutrients and sunlight on aquatic ecosystem function (Kisaka et al., 2021).

Streams and lakes were slightly acidic (pH < 6.0) with relatively low electrical conductivity (EC < 10.00 µS/cm), dissolved oxygen saturation (%OD) above 70%, and low nutrient concentrations (Appendix 1). Climatic seasonality also influenced aquatic systems, with higher temperature, turbidity, pH, EC, and % OD measured during the rainy months (Fonseca and de Mendonça-Galvão, 2014). While most streams were generally in pristine condition, for some streams we noted local signs of incipient water quality degradation. For example, both a stream with a recently established nearby rural area (‘Onça’) and a stream with margins facing urban influences (‘Gama’ stream) had higher nutrient concentrations, pH, and electrical conductivities compared to pristine streams. However, no increasing trends were observed (Appendix 1).

Both lotic and lentic systems in the AGCV had a high diversity of microinvertebrates (12 cladoceran species)-about 21% of the FD total richness in a unique lake (Sousa et al., 2018)-and macroinvertebrates (50 families and 28 genera of insects; unpublished data). Ephemeroptera, Plecoptera, and Trichoptera were the most common and frequent taxa in the community. Importantly, individual streams have unique community structures, with low similarity between streams. We also recorded several new occurrences of aquatic insect species for the FD: Macrelmis sp., Heterelmis sp., Cylloepus sp., Gyrelmis sp., and Xenelmis sp. (Elmidae, Coleoptera). Additionally, we recorded low population densities of the endemic species Macrobrachium candango (Crustacea), with no clear population trends over time.

The AGCV was sampled continuously for non-volant small mammals (rodents and marsupials) between 2008 and 2018. We sampled several plant physiognomies during this period, including typical savannas, savanna woodlands, and gallery forests. There was a total trapping effort of 223,698 trap-nights, which resulted in 4707 captured individuals. This massive effort in different physiognomies and distinct layers of the forests (including understory and canopy) allowed us to record 20 rodents and eight marsupials, representing 25% of the entire small mammal richness recorded in the Cerrado domain (Mendonça et al., 2018). Through these efforts, a previously undescribed species of a strictly arboreal spiny rat was captured and described for the first time (Phyllomys centralis, Echimyidae; Machado et al., 2018). Moreover, this continuous and long-term sampling effort advanced scientific knowledge of the theoretical and applied aspects of small mammal ecology and general biology, specifically related to long-term responses to fires, which are a significant disturbance in the Cerrado (Miranda et al., 2009). We found that small mammals of forested formations are not as resilient to fire disturbances as species from a typical savanna environment (Mendonça et al., 2015). In addition, fire in forests drastically altered habitat characteristics and favored the invasion of generalist rodents from open Cerrado habitats at the expense of more specialized forest species (Camargo et al., 2018). Thus, forest burning contributes to the biotic homogenization of mammalian fauna in the Cerrado. In addition, fire occurrence has potential effects on the mid- and long-term regeneration of forest trees. We observed that even three years after fire, the role of mammals as seed predators and potential dispersers is still affected, with a sharp reduction in seed removal of six native tree species in gallery forests. This indicates that the dispersal services provided by mammals (mainly the agouti Dasyprota azarae) for large-seeded species may be jeopardized by forest burning (Cazetta and Vieira, 2021). Such indirect effects of fire on ecosystem function could only be assessed through such a long-term research project.

Due to the high levels of habitat conversion and degradation, the Cerrado has been invaded by non-native species, some of which are particularly invasive (Zenni et al., 2018). Drosophilid fly assemblages have been monitored in the AGCV since 1998, revealing 80 neotropical and 11 non-neotropical species. Three of these exotic species were established in the Neotropical Region during this monitoring period. Zaprionus indianus, an Afrotropical species, was first detected in South America in 1998 (Vilela, 1999) and in our collections in 1999 (Tidon et al., 2003). This species is known as the fig fly because of its catastrophic impact on fig plantations, but is widely a generalist species and has been recorded breeding on 41 plant species (representing 25 families) in the Neotropical Region (Valadão et al., 2019). Drosophila nasuta, a species of Indo-Pacific origin, was first collected in the AGCV in December 2013 (one specimen). Two years later, its abundance reached 20% of drosophilids in this area (Leão et al., 2017). In 2015, D. nasuta was recorded in the Atlantic Forest of southeastern Brazil (Vilela and Goñi, 2015), and is also widespread in the northeast region of the country which is dominated by semiarid vegetation (Montes et al., 2021). The third newly arrived species, Drosophila suzukii, is known as the spotted wing Drosophila. It originated in southeast Asia but has gained pest status in Europe and North America as it damages many fruit crops. D. suzukii was recorded in southern Brazil in February 2013 (Deprá et al., 2014) and ten months later was collected in the AGCV (Paula et al., 2014). The early detection of these non-native species provided a unique opportunity to study their population dynamics in newly invaded communities (Döge et al., 2015; Mata and Tidon, 2013; Roque et al., 2013) and emphasizes the need to monitor cultivated areas in the region.

From 2017 to 2019, we sampled fruit-feeding butterflies (Lepidoptera, Nymphalidae) along a gradient of Cerrado physiognomies of the Brasilia Botanical Garden (JBB) ranging from woodland forests to typical savannas. We used a standardized protocol applying Van Someren-Rydon traps (Lucci Freitas et al., 2014), as this sampling scheme is low-cost and easy to sustain for long periods. We sampled twice a year: at the end of the rainy season (February–March) and at its onset (September–October). Four sampling stations, each containing four traps baited with bananas fermented in sugarcane juice were checked daily for eight consecutive days in each season. We photographed all captured butterflies with open wings, dorsally and frontally, and marked and released them immediately. We uploaded pictures and metadata containing the date, time, geolocation, and observations in an iNaturalist project (https://www.inaturalist.org/projects/monitoramento-de-borboletas-no-jbb).

We sampled 2561 butterflies with 768 traps*days of sampling effort for an average of 3.3 butterflies/trap/day. The butterfly assembly numbered 38 species, with three additional potential species added after a detailed revision of specimens from the genus Yphthimoides and the tribe Euptychiina. This number represents over a quarter of the 132 species registered at the Federal District (Emery et al., 2006) and is unlikely to increase significantly with additional effort as sampling locations are fixed under this sampling scheme. However, one of the most common species sampled at JBB—Nhambikuara cerradensis (Freitas et al., 2018)—was described after we initiated sampling, prompting us to revise the whole dataset to identify previously captured individuals. This event was an important test of the quality of our data registry choice, which allowed repeated verification of all observations.

We sought to define the proportion at which each species relates to an intact or disturbed environment, as this will allow us to establish a standard for monitoring changes in the butterfly community over time. Some species appear as possible indicators: Callicore sorana, Siderone galanthis, Eunica bechina, and Opsiphanes invirae. Additionally, the proportion of the two most abundant species (Hamadryas februa and Hamadryas feronia) seemed to change according to the degree of disturbance of the Cerrado areas.

In general, depth and width of all streams tended to decrease during the dry season, as a result of low water levels in aquifers. However, we observed more significant reductions in Onça stream volume. The headwater of Onça stream is the only one located in a small rural area, with agricultural activities. Moreover, Roncador lake has completely dry up during the extreme drought in 2017. From August to November 2017 (dry season), the volume of Roncador lake fell and recovery slowly just in 2018. However, from August to November 2018, there was a new period of below-average rainfall, resulting in a reduction in Roncador water level; however, a complete dry out was not observed again.

Climate variability with more extreme drought events is projected to continue to increase for the central part of Brazil in the future (South American Moon Region) (IPCC, 2021). Climate variability and changes in vegetation structure may interact to alter aquatic ecosystems. The low-order streams studied have a low resistance to disturbances, such as increased light incidence and nutrients, and showed significant differences in chlorophyll-a concentration (Clo-a) and ash-free dry matter (AFDM). More shaded streams maintained low Clo-a values, even in areas with higher nutrient concentrations. The density of macroinvertebrate scrapers also increased significantly with sunlight exposure and nutrient enrichment, as a response to biofilm increase (Kisaka et al., 2021). This is an essential indication of the low resilience of small pristine streams to the loss of the riparian zone and nutrient intake from fertilizers. Changes in riparian gallery forest cover and an increase in water nutrients can affect the characteristics and functioning of food webs in these streams.

We took advantage of this long-term research project to evaluate the potential effects of several sources of variation on the population dynamics and health condition of small free-living mammals, including possible harm caused by the occurrence of parasites. To this end, we provided food supplements to small mammals in their natural environments and monitored their responses for seven years. We observed that food supplementation does not directly affect the density of the most abundant rodent (long-tailed climbing mouse Rhipidomys macrurus) or the health (evaluated by changes in hemoglobin concentration) of the marsupial Gracilinanus agilis. However, climatic variables (as measured by maximum daily temperature and humidity) but not seasonality or food supplementation negatively affects hemoglobin concentration in animals parasitized by warble bot flies (Cuterebra apicalis) (Zangrandi et al., 2019). Thus, dehydration appears to interact with parasite-induced anemia, causing the worst health conditions during the driest days. We concluded that bot fly parasitism might become a significant threat to small populations considering projected increases in the frequency of drought conditions due to climate variability. Moreover, climate variability may also affect small mammal communities. We detected a trend of declining open-area species density (mainly in wet grasslands) over time, which requires further investigation in future studies.