Although Amazon provides crucial goods and ecosystem services for humanity, the alarming conversion of its original ecosystem due to agricultural activities challenges the maintenance of biodiversity in this rainforest. Beetles from the family Melolonthidae serve as a proxy for assessing the effect of land use change on biodiversity. In this study, we evaluated the effect of land use changes on the entire Melolonthidae community (species richness, abundance, and biomass) and on each subfamily (Rutelinae, Melolonthinae, and Dynastinae). Peach palm Bactris gasipaes Kunth (Arecaceae) plantation, coffee Coffea canephora Pierre ex A. Froehner (Rubiaceae) plantation, pasture Urochloa decumbens Stapf. (Poaceae), Forest remnants had their Melolonthidae communities assessed in a southern Amazon region. Forest remnants encompassed a higher species richness of the Melolonthidae and Rutelinae compared to the other habitats, while pasture areas had a higher abundance of Melolonthidae, Dynastinae, and Melolonthinae, as well as a higher biomass of the last group. Coffee plantation presented the lowest species richness, abundance, and biomass for all taxonomic levels studied. The results provide evidence that the Melolonthidae beetle community, especially Rutelinae, is negatively affected by land use change in the Amazon, reinforcing the role of these beetles in conservation studies. Pasture areas are favorable habitats for some species of the Dynastinae and Melolonthinae subfamilies, which suggests that they may be pests in this ecosystem. Future population studies are necessary to broaden our understanding of this group of insects.

Study area

This study was carried out in Juína municipality, which is located in the Amazonian ‘arc of deforestation’, in the state of Mato Grosso, Brazil (11°25’ S; 58°46’ W; 320 m a.s.l., see Figures 1a and 1b). The climate of the region is a transition between tropical monsoon (Am) and tropical warm and humid (Aw) according to the Köppen classification (Alvares et al., 2014). The region has ca. 2,000 mm of mean annual rainfall and a mean temperature of 24 °C, ranging between 20 and 40 °C (Batistão et al., 2013). There are two marked seasons: the dry season (from May to September, with less than 100 mm of mean monthly rainfall) and the rainy season (from October to April, with ca. 330 mm of mean monthly rainfall (Alvares et al., 2014). The dominant physiognomy is terra firme, a non-floodable evergreen rainforest; besides terra firme, there are floodable patches of tropical rainforest (várzea).

Therefore, this study was performed in a conserved forest and three other habitat types representing the non-native land use types of the Juína Amazonian region: peach palm- and coffee plantations and pasturelands (Figure 1c).

(i) Conserved forests are forest fragments ranging from 78 to 790 ha that are located in conservation units. They had minor signs of anthropogenic activities (e.g. logging) and common Amazonian plant species, such as Bertholletia excelsa Bonpl., 1808 (Lecythidaceae), cedro Cedrela fissilis Vell., 1829 (Meliaceae), sumaúma Ceiba pentandra Gaertn., 1791 (Malvaceae), copaíba Copaifera glycycarpa Ducke, 1933 (Fabaceae), tamburil Enterolobium contortisiliquum (Vell.) Morong, 1893 (Fabaceae), ipê-amarelo-da-mata Handroanthus serratifolius (Vahl) S.O.Grose, 2007 (Bignoniaceae), jatobá Hymenaea courbaril L., 1753 (Fabaceae) and ingá-cipó Inga edulis Mart., 1837 (Fabaceae).

(ii) Peach palm (B. gasipaes) plantations were established between 2012 and 2016, and such habitats range from 0.5 to 5 ha. Peach palm plantations comprised plants set 1 m apart from each other in line transects with 2 m between lines, resulting in a density of 5,000 peach palm plants/ha. Each plant reached about 3 m in height, resulting in a closed-canopy vegetation structure. Palm heart extractivism of peach palms occurs when plants reach ca. 1.70 m of height. Such extraction is performed every four months, and after cutting, the producer generally manages thinning so that there are only three vigorous tillers per clump.

(iii) Coffee plantations encompassed the conilon/robusta species (C. canephora), and their areas ranged between 2.1 to 9.2 ha. Plants spaced 3 × 2 m between them, which resulted in a density of 1,666 coffee plants/ha. Land management was performed to control plant pests and diseases, as well as to fertilize soils.

(iv) Pasturelands comprised the exotic pasture U. decumbens set from 8 to 22 years before this study. This habitat has areas ranging from 4.5 to 10 ha, and no fertilizer has been applied to these pasturelands. Cattle that inhabit these pasturelands are vaccinated against aphthous fever and brucellosis, and ivermectin, doramectin, and oral vermifuges (albendazole) are also used to control cattle parasites.

Beetle sampling

Melolonthidae beetles were sampled monthly between November 2021 and October 2022. Each habitat type (conserved forest, peach palm, coffee plantations, and pasturelands) was sampled in four distinct sites, spaced at least 1 km between them, totaling 16 replicates. Detailed information on each of the 16 replicates, such as coordinates and size in hectares, can be found in Table S1. This study had an effort of 192 samples (12 months × 4 habitat types × 4 sites).

One light trap (rechargeable 100 W LED light) was set in the center of each of the 16 study sites to sample Melolonthidae. These light traps were adapted from “Luiz de Queiroz” traps (Figure S1), being successfully used to attract Melolonthidae beetles (Buss, 2006; Aragón-García et al., 2008; García-López et al., 2011). Traps were left in the field for 12 hrs – between 18:00 and 6:00 of the following day. The insects were deposited in plastic bags with alcohol 70% and subsequently sorted in the biology laboratories of Instituto Federal de Ciência, Educação e Tecnologia de Mato Grosso (IFMT), Campus Juína.

Melolonthidae beetles were identified at the genus level with the aid of taxonomic keys (Cherman & Morón, 2014; Cherman et al., 2019; Duarte & Grossi, 2024; Evans, 2003; Frey, 1973; Krajcik, 2008; Morón, 1990), and species were identified with the aid of one of the coauthors (FZVM). Individuals were deposited in the Entomological Collection of Universidade Federal do Mato Grosso (CEMT – UFMT) (Cuiabá, MT, Brazil) and in the Entomological Collection of IFMT (Juína, MT, Brazil). To record the biomass of species, all collected individuals were dried (40 ± 5 °C) and weighed on a 0.0001 g precision balance.

Data analysis

To estimate the sampling efficiency of Melolonthidae in each habitat type, iNEXT's sampling coverage approach was used (Chao et al., 2014; Hsieh et al., 2016) with iNEXT online software. This is a measure of sample completeness and reveals the proportion of the total number of individuals in a community that belong to the species represented in the sample. (Hsieh et al., 2016).

Effects of habitat type on Melolonthidae assemblage

The following analyses were performed separately for the entire Melolonthidae assemblage and for each collected subfamily (Dynastinae, Melolonthinae, and Rutelinae). The Sericinae subfamily was not analyzed separately due to the limited data for this group (see Supplementary Material, Table S2). Generalized Linear Mixed Models (GLMMs) were used to assess the effect of habitat types (conserved forest, peach palm- and coffee plantations, pastureland – predictor variables) on Melolonthidae species richness, abundance, and biomass (response variables). Habitat type was considered the fixed factor, and the months of collection were considered the random factor in the model. To select the best-suited models, residual analyses were tested on GLMMs (Crawley, 2013). After that, contrast analyses were performed to test the difference between pairs (Crawley, 2013). GLMM analyses were done in RStudio software, version 4.3.3 (R Core Team, 2023).

The effect of habitat type on Melolonthidae assemblage structure (i.e., species distribution and abundance) was tested with Permutational Analysis of Variances (PERMANOVA). This statistical approach was run with 999 permutations (Anderson, 2001). In order to test the differences in the multivariate dispersion of data from each habitat type, Permutational Multivariate Analysis of Dispersion (PERMDISP) (Anderson, 2001) was used. The PERMDISP results are reported as the average within-group distance of each group point from its centroid. Detection of statistical differences in inter-group variability can be used as a viable test for changes in beta diversity between habitat types (Anderson 2006; Anderson et al. 2006). To visually explore the groupings of Melolonthidae data and each subfamily separately, Non-Metric Multidimensional Scaling (NMDS) was used (Anderson & Willis, 2003). PERMANOVA, PERMDISP, and NMDS were performed based on the Bray-Curtis similarity matrix, which is sensitive to species abundance data. Data were square-root transformed in square root to avoid bias that could be caused due to the superabundant species. PERMANOVA, PERMDISP, and NMDS were done in Primer software, version 6.0 (Clarke & Gorley, 2006).

Indicator Value (IndVal) was used to determine the species indicator potential of each studied habitat (Dufrêne and Legendre, 1997). IndVal considers species fidelity and relative distribution in each treatment (i.e., habitat type). In this study, 5,000 randomizations were performed to estimate the indicator value of each Melolonthidae species (Monte Carlo test; p < 0.05). IndVal was performed in the ‘indiespecies’ package in R Studio version 4.4.3  (Cáceres et al., 2022; R Core Team, 2023).