Sleeping accommodations for researchers increase the likelihood of biodiversity inventories in protected areas

Biodiversity assessments are important for successful conservation initiatives. However, many protected areas (PAs) still have no biodiversity inventories of important species‐rich taxa, such as invertebrates, that deliver important ecosystem services (e.g., pollination, nutrient cycling) or are biological indicators. Here, we investigated if the likelihood of having arthropod inventories in PAs (state parks) was influenced by PA size, date of PA creation, distance to the nearest university, main vegetation type (forest/savanna), and availability of accommodations for researchers by inspecting management plans from 61 PAs in southeastern Brazil. Only 13 PAs (22%) had arthropod inventories considered in park management plans. The availability of accommodations for researchers was the only variable influencing the likelihood of a PA to have arthropod inventories. A modest investment (around US$35,000 in Brazil) is required to make accommodations available for researchers in a PA and may positively impact information acquisition that is key to planning and protecting biodiversity.

Plain language summary Successful conservation initiatives usually require previous information available owing to basic biological data acquisition, such as those obtained by biodiversity inventories. However, many areas of the world are still undersampled for many taxa, including Protected Areas (PAs). We tested the influence of several variables (distance from the closest University and city, PA size, main vegetation type, date of PA creation, and availability of accommodations) in the likelihood of arthropod inventories in 61 PAs from southeastern Brazil. Only the availability of accommodations affected positively the chance of a PA to have an inventory of arthropods. Our model suggests that providing simple accommodations for researchers may increase the chances of biodiversity assessments. Funds to make accommodations available may come from taxes, donations, and support from private companies affected by arthropod-delivered services/disservices. If costs to build and maintain accommodations are out of reach, options such as trailer/motorhome rentals may be considered. Since arthropods can be used as proxy indicators of habitat quality and availability for other organisms, provisioning of accommodations may also benefit information acquisition and conservation of other taxa as well.

| INTRODUCTION
Data about biodiversity distribution are critical for conservation planning and action (Oliveira et al., 2017). Changes in land use and global climate have an enormous influence on the occurrence and abundance of species, and these effects are likely to increase in the coming decades (Pereira et al., 2010). However, there are still large gaps in detailed information on species occurrence, abundance, and distribution for many groups of animals and plants around the world (Hughes et al., 2021). Data availability is often influenced by accessibility to sampled locations (proximity to cities, roads, and rivers), distance from experts (from universities, museums, and botanical gardens), and the occurrence of Protected Areas (hereafter, PAs) (Hughes et al., 2021;Moerman & Estabrook, 2006;Oliveira et al., 2017;Ribeiro et al., 2016). Researchers might be attracted to larger PAs because these are likely to be more heterogeneous. Larger PAs are also more likely to be hit by biodiversity surveys than smaller PAs if surveys are done in systematic designs at large spatial scales (Thomas et al., 2004). Older PAs are probably more likely to have been sampled than new ones considering an increasing cumulative chance of a survey with time. The availability of sleeping accommodations in PAs may provide facilities for researchers to perform sampling during the day and night, prepare food and rest between sampling sessions, and potentially bias the likelihood of inventories compared to remote locations where there are no accommodations available. Vegetation type may also bias the surveys, as forests are usually perceived as more valuable, and as such, more studied than savannas and grasslands (Parr et al., 2014).
The scarcity and biases of information impose difficulties in monitoring the trends in populations/ species targeted for conservation and force conservation policies to rely on extrapolations of the data available and modeling approaches (e.g., species distribution models) (Hughes et al., 2021;Oliveira et al., 2017). Even these approaches, however, are sensitive to data availability (e.g., number of sites surveyed) and quality (e.g., proper taxonomic identification, and accurate location of records), increasing the uncertainty about the results and sometimes leading to misleading conclusions and recommendations for conservation (D'Amen et al., 2013;Delso et al., 2021). The challenges to cope with scarce and often poor-quality information are even higher for highly diverse groups, such as invertebrates. For instance, while pervasive declines in fauna have been reported in the last decades, the uncertainty (variance) around estimates of declines in species diversity and abundances are much higher for insects compared to vertebrates (e.g., birds and mammals) (Dirzo et al., 2014;Wagner et al., 2021).
Most of the tropics are still undersampled for several taxonomic groups, although the data available clearly indicate that the tropics hold the largest diversity of animals and plants on Earth (Hughes et al., 2021). The tropics also hold the highest current rates of deforestation and land use changes (Haddad et al., 2015), increasing the threats faced by biodiversity and the pressures for effective measures to reduce or mitigate the current losses. The creation of PAs is a long recognized and effective initiative to protect local biodiversity (Bruner et al., 2001). However, many PAs are created based on charismatic species (usually vertebrates), scenery and landscape features, and often on partial knowledge of the local biodiversity, which may lead to decisions that are biased and misguided according to conservation priorities (Delso et al., 2021). Several invertebrates may already be responding to climate change, which has the potential to change the abundance and spatial distribution of many species and their habitats (D'Amen et al., 2013;Hallmann et al., 2017;Wagner et al., 2021). However, without detailed information about biodiversity distribution in space and time, conclusions may be elusive (Wagner et al., 2021). Therefore, an increase in our understanding of how biodiversity is distributed in tropical ecosystems in general, and PAs in particular, is critical if we intend to mitigate the diverse threats faced by biodiversity (Oliveira et al., 2017). A better knowledge of how biodiversity is distributed may help to implement effective measures of protection and adaptability to land use and climate change.

Practitioner points
• Biodiversity assessments are needed to better understand insect declines worldwide. • Arthropod inventories are more likely if accommodations are available in Protected Areas. • Providing accommodations for researchers may increase the likelihood of biodiversity assessments.
potentially critical for biodiversity and ecosystem service conservation, as well as in pest control and monitoring of pandemics. Recent reports of massive declines in diversity and abundance of many insect taxa, even inside PAs, highlight the need for biodiversity inventories and monitoring at more sites in the long term, as the reasons behind several declines are not entirely known (Hallmann et al., 2017;Lewinsohn et al., 2022;Wagner et al., 2021; but see Crossley et al., 2020).
Here, we investigated some potential factors influencing the likelihood of biodiversity assessments of arthropods in 61 PAs (state parks) from southeastern Brazil. Since post-graduate programs provide the bulk of biodiversity science in Brazil, we tested how distance from Universities with post-graduate programs in Zoology, Entomology, or Ecology, distance from the nearest city, PA size and date of PA creation, vegetation type, and the presence of accommodations for researchers may influence the chance of a PA having arthropod surveys.

| METHODS
Our study covered all state parks established in São Paulo state (248,000 km 2 , roughly the size of the United Kingdom) in southeastern Brazil. We surveyed (in November 2021) PAs specifically devoted to the protection of biodiversity under the administration of an environmental state agency. Together, these PAs (N = 61) cover an area of ca. 1 million ha (Fundação para a Conservação e a Produção Florestal do Estado de São Paulo, 2022). A list of the PAs surveyed is available in Supporting Information: Table S1.
For each PA we retrieved the information available in the management plan, an official document that provides a detailed review of all information available about the PA, and that is used to guide local conservation and management policies (Brasil, 2000). For those PAs that still has not approved a management plan, we consulted the park manager directly by email in search of information. As potential explanatory variables for the existence of biodiversity inventories, we recorded the position of the PAs (Lat-Long), the linear distance to the nearest city, and the nearest University Campus with a post-graduate program in Zoology, Entomology, or Ecology (Supporting Information: Table S1). We also recorded the main vegetation type of the PAs (Cerrado or Atlantic forest, the dominant vegetation domains in the state), PA size (in hectares), date of PA creation, and the presence of accommodations for researchers. We considered Reserva Biológica de Mogi-Guaçu and Estação Ecológica (EEco) de Mogi-Guaçu (Supporting Information: Table S1) as a single sampling unit, as they border each other (i.e., with the perimeter of a PA in direct contact with the other PA), and researchers often use the accommodations available in one of the parks to access sites of the other park. Keeping both as independent observations would create a pseudoreplication for data analysis. Therefore, the final sample size for statistical purposes is N = 60 PAs. Although the PAs EEco Assis and EEco Paranapanema had no in situ accommodations, both are bordered by public institutions with accommodations available for researchers (Floresta Estadual de Assis and Floresta Estadual de Paranapanema). Therefore, we assigned accommodations available to both.
To test the influence of the predictor variables PA size, year of PA creation, distance from the nearest city, distance from the nearest University Campus, main vegetation type, and the presence of accommodations in the PA (fixed variables) on the presence/absence of arthropod biodiversity inventories (response variable), we fitted a generalized linear model (GLM) using a binomial distribution (Zuur et al., 2009) and glm function in R (R Core Team, 2019). We started with the full model and proceeded with model simplification by subsequent removal of the term with the highest p value among the nonsignificant ones followed by model comparisons with a likelihood ratio test with the aid of anova function. Residual diagnostics were performed with the aid of simulateResiduals function in DHARMa package (Hartig, 2022). Residuals did not show an increase in variance across the range of fitted values; therefore, the models were considered valid. To obtain the proportional reduction in deviance explained by our model, we calculated a pseudo-R 2 (R L 2 ) (Cohen et al., 2002) given by a are the residual deviances obtained, respectively, by fitting the response variable to a null model (i.e., a constant equal to 1) and to the model with the explanatory variables selected after model simplification as described above.

| RESULTS
Most PAs surveyed had management plans, either in Cerrado (8 in 10; 80%) or in the Atlantic forest (33 in 50; 66%) (Supporting Information: Table S1). For eight PAs, information about the availability of accommodations and arthropod surveys was obtained from direct contact with park staff. We did not find management plans or information about the presence of accommodations for 19 and 11 PAs, respectively (Supporting Information: Table S1). PAs started to be created in 1896 in the state, covering areas ranging from 17 ha to 332,000 ha each. PAs were from 5 to 344 km distant from the nearest University Campus, and from 1 to 108 km from the nearest city. The presence of accommodations was the only predictor variable that influenced the likelihood of a PA to have arthropod inventories (Binomial GLM: estimate 2.262 ± 1.094, z = 2.07; p = 0.038; pseudo-R 2 = 0.20; Figure  1). This relationship holds even if we excluded all PAs without management plans from the analysis (Supporting Information). Considering all PAs from where we could retrieve the information about the presence of accommodations and arthropod surveys (N = 49), 12 PAs had accommodations for researchers and biodiversity inventories, while 20 PAs had accommodations but not inventories, 16 had no accommodation and no inventories, and only 1 had no accommodations but had an inventory of arthropods ( Figure 1 and Supporting Information: Table S1).

| DISCUSSION
Our findings indicate that distance to the nearest University Campus, distance to the nearest city, PAs size, date of PA creation, and main vegetation type do not explain the presence of arthropod biodiversity assessments in PAs from southeastern Brazil. However, we found that PAs with accommodations for researchers are more likely to receive a survey of arthropods than PAs with no availability of accommodations ( Figure 1). These results suggest that the provisioning of simple accommodations is likely to attract researchers and increase the likelihood of biodiversity inventories that are key to successful conservation initiatives in an era of global change (Eggleton, 2020;Oliveira et al., 2017). Since arthropods can be used as proxy indicators of habitat quality and availability for other organisms (Gerlach et al., 2013), the provisioning of accommodations may also benefit information acquisition and conservation of other taxa as well.
Although we did not investigate why accommodations for researchers increase the likelihood of biodiversity inventories, we can suggest several reasons. For instance, even simple accommodations may provide facilities where researchers can organize, protect, and store equipment during field sampling. Accommodations may also provide a place where researchers may sort, label, and prepare samples for proper identification and preservation in collections, enhancing the quality of samples for further work. Some arthropods may have delicate organs or body parts that are important for identification and that can be easily damaged if not preserved properly in a short time after sampling (Santos & Fernandes, 2021). If a facility is available, researchers may anticipate some stages of work and avoid transporting samples for long distances, decreasing the chance of damaging samples. Accommodations may also provide better logistical conditions for researchers that need to stay for long periods or to access distant areas, allowing them to rest after sampling, and enhancing their welfare. Some arthropod groups are also more easily sampled at night (Callisto et al., 2021), and the presence of accommodations is likely to benefit such sampling schemes. PA accommodations are sometimes the only option for accommodation in the surroundings and tend to be cheaper compared to private accommodations, an important aspect in a time of budget cuts for research (Kowaltovski, 2021;Levis et al., 2020). The presence of accommodations may be more favorable for studies applying certain sampling methods/ designs, such as inventories embracing many days to be accomplished or that require samples at night. Accommodations may also allow more intense and repeated sampling in PAs. These factors may be further explored in future studies.
Although the availability of accommodations is important, not all PAs with accommodations available have arthropod inventories. These results are in part probably due to a number of hard-to-control and unmeasured factors, attested also by the relatively low pseudo-R 2 (0.20) obtained for our explanatory model. For instance, accommodations may be available but building conditions may be poor and a PA may have no regular maintenance team due to a shortage of funds and staff (authors's personal observation). Some PAs may be in unsafe spots or sites in conflict with local communities that sometimes do not behave friendly to PAs staff and researchers (De Pourcq et al., 2017). All these may decrease the chance that researchers sample in these PAs.
Financial support and taxonomy experts, among others, are also probably critical factors. Not surprisingly, the best inventoried places in the world are those in wealthier nations, with better infrastructure for research and taxonomic collections, and regularly financed research teams (Hughes et al., 2021). Strategic funds and fellowships directed to biodiversity surveys may help to overcome some of these constraints in less surveyed tropical countries (Noll et al., 2022). On the other hand, some PA staff may be more active in searching for collaborations with research institutions than others. Direct contact between PAs staff and academics may increase the attractiveness of researchers to perform basic or applied research (Durigan et al., 2007), and the occurrence of specific taxa or threatened species F I G U R E 1 Arthropod inventories are disproportionally more common in protected areas (PAs) having accommodations available for researchers than PAs without accommodations in southeastern Brazil (binomial generalized linear model: estimate 2.262 ± 1.094, z = 2.07; p = 0.038). may enhance the likelihood of partnerships between research institutions and PAs staff (e.g., Instituto de Pesquisas Ecológicas, 2022). Therefore, although we found evidence that accommodations may increase the likelihood of biodiversity surveys, they should work better in combination with some local policies and good practices to enhance biodiversity samplings in PAs.
A modest investment is required to provide simple accommodations for researchers in PAs. For instance, from price quotations with specialized companies, a prefab wooden house with 90 m 2 , two rooms, a shared bathroom, a kitchen and basic furniture (two bunk beds, a stove, a refrigerator, a table, and five chairs) may be built for around US $35,000. Depending on the local availability of materials and labor force, these prices may be lowered. Other options may include trailer/motorhome rental for places where infrastructure or local funds would not be easily available. Funds to make accommodations available may come from taxes and environmental compensatory funds (Brasil, 2018;São Paulo, 1993), donations and support from private companies affected by arthropod-delivered ecosystem services/disservices (Schowalter et al., 2018). Unfortunately, PAs and environmental research initiatives have been dealing with increasing budget cuts in recent years in Brazil (Levis et al., 2020). Our results suggest that even modest increases in funding directed to make accommodations locally available may have the potential to enhance the ability of parks to acquire information about tropical biodiversity, helping them to better plan and act to protect it. For PAs that already have accommodations available but no biodiversity surveys, regular inspections and actions to improve accommodation conditions combined with contact with researchers may point out potential solutions to enhance the likelihood of inventories.

ACKNOWLEDGMENTS
We thank all managers of Protected Areas who attended our invitation to provide data about the availability of accommodations and arthropod surveys. GPF thanks Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) for an undergraduate scholarship (iniciação científica; PIBIC/CNPq/UFSCar). MLC thanks Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES; proc. 88887.505779/2020-00) for a scholarship. MLC and AVC thank Neotropical Grassland Conservancy and The Rufford Foundation for research grants. We dedicate this article to all hard workers from Fundação Florestal who dedicated their lives to protecting our biodiversity in an era of environmental protection defunding and science denialism. This study was supported by Ministério da Ciência, Tecnologia e Inovação, Conselho Nacional de Desenvolvimento Científico e Tecnológico, Coordenação de Aperfeiçoamento de Pessoal de Nível Superior 88887.505779/2020-00, Neotropical Grassland Conservancy and The Rufford Foundation.