Bats constitute about 22% of known mammal species; they have various ecological roles and provide many ecosystem services. Bats suffer from several threats caused by anthropization, including exposure to toxic metals and metalloids. In our systematic review, we analyzed 75 papers to investigate how species, diet, and tissue type impact bioaccumulation in bat species all over the world. Most studies documented element accumulation in fur, liver, and kidney; at least 36 metals and metalloids have been measured in bat tissues, among the most studied were mercury and zinc. Comparisons with known toxicological thresholds for other mammals showed concerning values for mercury and zinc in bat hair, lead and some essential metals in liver, and iron and calcium in kidneys. Moreover, accumulation patterns in tissues differed depending on bat diet: insectivorous bats showed higher metal concentrations in fur than in liver and kidney while frugivorous species showed higher values in liver and kidney than in fur. This review points out several information gaps in the understanding of metal contamination in bats, including a lack of measured toxicity thresholds specific for bat tissues. Data on trace element bioaccumulation and its associated health effects on bats is important for conservation of bat species, many of which are threatened.

The literature search was performed using Google Scholar and Web of Science Core Collection (Clarivate) during the period of 1978 to 2021, using keywords/Boolean operators: “bats” and/or “metal” and/or “contamination” and/or “fur” (title). A total of 75 publications from the 1978-2021 period were located and included in this review, including peer-reviewed papers and unpublished reports. Metadata collected from 67 scientific papers (reviews excluded) included: species, sample size, tissue, geographic location, age, sex, diet type, and trace element content. Element concentrations in bat tissues were converted from a wet weight to a dry weight by multiplying the wet weight value by a factor of 3.7 for liver, 3.9 for kidney, and 4.0 for fur and all the other analyzed tissues; these factors were calculated according to the moisture content of the specific organ. No such conversion was made when dry weights were provided. Data analysis included: metals (Ag, Al, Ba, Be, Ca, Cd, Cr, Co, Cu, Fe, Hg, K, Li, Mg, Mn, Mo, Na, Ni, Pb, Rb, Sn, Sr, Th, Ti, Tl, V, W, Y, Zn, Zr), metalloids (As, Si), and nonmetals (Ar, Br, O, Se), in various tissues (fur, liver, kidney, whole body, bone, brain, blood, stomach, muscle, heart, wing, lungs, patagium, spleen, gut), and excrement (guano).  Tissue burdens of trace elements were compared among the bat diet categories (insectivores, frugivores, nectarivores, and other). All trace element concentrations for bats provided in every article were pooled by element, and log-transformed prior to statistical analysis. Four bat species (Eptesicus fuscus, Myotis daubentonii, Myotis lucifugus, and Myotis myotis) have been subject of multiple studies; it was possible to perform comparative statistical analyses on tissue concentrations among these species.

A meta-analysis of published data was carried out for metal concentrations among bat tissues, species, and diet categories. The following statistical analyses were performed.

1. Normality tests (Kolmogorov-Smirnov, Shapiro-Wilk, Anderson-Darling, Lilliefors, and Jarque-Bera) were conducted for each data set (i.e., for every metal in each tissue, species, and diet category), to establish whether data distributions were normal. Significance was set at p = 0.05.

2. For normally distributed data, a parametric test (One-way ANOVA) was performed to compare each element’s concentration among different tissues, species, or diet categories, followed by a Tukey test (with a confidence interval of 95%) and an REGWQ test to determine the specific differences among the categories.

3. For non-normally distributed data, a Kruskal-Wallis test was applied to compare each metal’s concentration between tissues, species, or diet categories, followed by a Dunn’s procedure to compute multiple pairwise comparisons, and a Bonferroni correction to avoid redundancy bias.