To perform a systematic quantitative review on SRE patterns we used different approaches, following the synthesis guidelines in Haddaway et al. (2020). First, we conducted literature searches in the Web of Science and Scopus searching engines, using the following keyword combination: (“altitud* gradient*” OR “elevation* gradient*” OR “mountain* gradient*” OR altitud* OR elevation* OR mountain*) AND (“species richness” OR richness OR diversity OR “species diversity" OR biodiversity). Initial searches were conducted between July 2020 and March 2021 and covered the timespan of the bibliographic databases (1945-2021). To complement our database, we also included studies provided in the reference list of elevational gradients reviews (e.g., Rahbek, 1995; Sanders & Rahbek, 2012; see Figure S1 and Table S1). A total of 644 published studies were fully screened for eligibility. 

From these, we selected all studies that reported data on SRE relationships. This search comprised global data across all biogeographical realms, all continents, except Antarctica, and insular regions. We focused on the dominant components of terrestrial ecosystems by reviewing SRE models reported for microorganisms, invertebrates, vertebrates, and plants. Studies were included when meeting the following criteria: 1) raw data on species richness (neither rarefied nor transformed) provided for each elevational gradient; 2) any statistical parameter stated for species richness along an elevation gradient (e.g., r, F, Spearman-rho, t or R2); and 3) reported sample sizes. In less than 5% of the database, we contacted the authors to obtain original raw data and calculated the correlation coefficients whenever missing from the original publication. Studies that treated elevation as a categoric or qualitative variable were excluded. This selection criteria resulted in 143 studies and 479 gradients that were screened for quantitative data. 

As a conservative approach, we further excluded the gradients reporting sample size lower than five elevational quotas sampled over a single mountain (n=2), undefined SRE models (i.e., no clear pattern reported in published figures or results, n=15 gradients, 3.13% of total cases), infrequent SRE models (e.g., polynomial models, n=14, 2.92% of total cases), and studies focusing exclusively on microorganisms (n=12, 2.5%), due to an insufficient number of records. Further, we excluded gradients with inconsistent sampling throughout the mountain and/or those with inflated sample sizes, likely due to pseudoreplication (n=63, 13% of records). Thus, the final database comprised 373 gradients from 127 studies comprising the three most frequent SRE models reported in the literature: linear negative, hump-shaped, and linear positive.