**2.2 Brazil nut production data**

The information regarding Brazil nut seed (**Figure 2**) production was obtained from sources at the Brazilian Institute of Geography and Statistics [6]. IBGE conducts annual data collection through the administration of questionnaires in each municipality within the study area. IBGE gathers this information by consulting agricultural

**Figure 2.** *Seeds of* Bertholletia excelsa *(Brazil nut). Photo source: Diego Oliveira Brandão.*

*Threats and Sustainability of Brazil Nut* (Bertholletia excelsa *Bonpl.*) *Pre-Industrialization... DOI: http://dx.doi.org/10.5772/intechopen.113715*

and livestock establishments, industries, companies, government agencies, and other organizations involved in Brazil nut extraction, using the reference year as the survey's base. The dissemination of data is carried out through the IBGE Automated Recovery System (SIDRA), under the Production of Vegetable Extraction and Forestry (PEVS) survey, Table 289, and is accessible openly and online. The information regarding the quantity of Brazil nut seed production is expressed in tons. The data used covers the period from 1990 to 2021 and was collected for all mesoregions within the study area (**Figure 1C**).

#### **2.3 Climatic variable data**

Monthly climate data for precipitation, relative humidity, and temperature (average and maximum) were obtained from the Meteorological Database (BDMEP) of the National Institute of Meteorology of Brazil [29] to represent the climatic variations in the Baixo Amazonas mesoregion. These climate data originated from meteorological station 82,178, located in the municipality of Óbidos, Pará, Brazil, in the Baixo Amazonas mesoregion. This station is at an altitude of 54.7 meters, latitude −1.905° S, and longitude −55.52° W, and operated from January 1927 until July 2021, currently being closed. The data covered the period from 1989 to 2020 and were requested on July 24 and August 28, 2023.

The monthly climate data was transformed into annual climate data using the arithmetic mean. This calculation was done by summing the monthly values and dividing the result by 12 (the number of months in a year). However, in some cases, monthly climate data for certain variables were not available from the BDMEP. To fill in these missing data, the monthly arithmetic mean for the period from 1989 to 2020 was used. However, it's important to mention that the maximum and average temperature data for the years 2019 and 2020 were also unavailable, and it wasn't possible to fill them with the arithmetic mean. The missing data represented 4% of precipitation data, 2% of relative humidity data, and 2% of temperature data (average and maximum).

Finally, annual climate data for Vapor Pressure Deficit (VPD) was obtained. However, INMET did not provide VPD data at the Óbidos station [29]. Therefore, the VPD calculator [30] from the College of Agriculture and Life Science at the University of Arizona was used to obtain annual climate data for VPD from annual climate data for relative humidity and average temperature obtained at the Óbidos station. VPD data is a climatic variable used to assess plant water stress [31]. MS-Excel software was used to process the climate data.

#### **2.4 Data analyses**

Analyses of variations in Brazil nut production were conducted for the Brazilian Amazon and the 26 mesoregions studied between 1990 and 2021 and are presented graphically. These results were also presented with statistics including minimum, maximum, arithmetic mean, and standard deviation. All the graphical figures presented were created by the authors using the software MS-Excel, R, and QGIS.

The Baixo Amazonas mesoregion (PA) was selected as the case study to assess the impact of climatic variables on the temporal fluctuations in Brazil nut production. This region, besides being one of the largest Brazil nut producers in the Amazon, also provides long-term time series of climatic data. Therefore, simple linear regression analyses were conducted to investigate the relationship between Brazil nut production and each of the five climatic variables.

The regression analyses were conducted using the climatic data from the same year in which the Brazil nut production data were recorded (referred to as the 'base year'). Additionally, climatic data from 1 year before seed collection were correlated with the seed production observations in the base year. This allowed for the assessment of potential delays in the response of Brazil nut production to annual climatic variations. Before conducting these analyses, the Shapiro-Wilk test was performed to check the normality of the seed production data, including residual analysis.

The values of intercept, slope coefficient, adjusted R-squared, and p-value estimated by the models were presented in the figure depicting the relationship between seed production and climatic variables. The intercept and slope coefficient represent the model's fitting parameters. The R-squared value and p-value indicate the explanatory power of the model and its significance. Hypothesis testing and model construction were conducted using the statistical software R, version 4.3.1 [32].
