Extinction risk of the world's freshwater mammals

The continued loss of freshwater habitats poses a significant threat to global biodiversity. We reviewed the extinction risk of 166 freshwater aquatic and semiaquatic mammals—a group rarely documented as a collective. We used the International Union for the Conservation of Nature Red List of Threatened Species categories as of December 2021 to determine extinction risk. Extinction risk was then compared among taxonomic groups, geographic areas, and biological traits. Thirty percent of all freshwater mammals were listed as threatened. Decreasing population trends were common (44.0%), including a greater rate of decline (3.6% in 20 years) than for mammals or freshwater species as a whole. Aquatic freshwater mammals were at a greater risk of extinction than semiaquatic freshwater mammals (95% CI –7.20 to –1.11). Twenty‐nine species were data deficient or not evaluated. Large species (95% CI 0.01 to 0.03) with large dispersal distances (95% CI 0.03 to 0.15) had a higher risk of extinction than small species with small dispersal distances. The number of threatening processes associated with a species compounded their risk of extinction (95% CI 0.28 to 0.77). Hunting, land clearing for logging and agriculture, pollution, residential development, and habitat modification or destruction from dams and water management posed the greatest threats to these species. The basic life‐history traits of many species were poorly known, highlighting the need for more research. Conservation of freshwater mammals requires a host of management actions centered around increased protection of riparian areas and more conscientious water management to aid the recovery of threatened species.


INTRODUCTION
Freshwater ecosystems occupy only 0.8% of Earth's surface but are home to approximately 10% of all known animal species (Balian et al., 2008) and one third of all vertebrates (Dudgeon et al., 2006).Freshwater ecosystems are one of the most threatened.Freshwater vertebrate populations have declined by 81% since 1970, and since 1900 an estimated 71% of the world's wetlands have been lost (Davidson, 2014;McRae et al., 2017).Key threats include flow modification, water extraction, habitat destruction and degradation, invasive species, and climate change (Collen et al., 2014;Dudgeon et al., 2006;Reid et al., 2019;Strayer, 2010).Pollution of waterways from domestic, industrial, and agricultural waste products is a leading cause of freshwater system degradation in regions densely populated by humans (Collen et al., 2014;Stoett et al., 2019).Climate change represents an even greater emerging threat to freshwater systems due to more severe drought from lack of rainfall, increased aridity, and greater frequency of flooding events (Dudgeon et al., 2006;Larkin et al., 2020).
One group that is often overlooked when discussing impacts on freshwater systems is freshwater mammals.This is a highly diverse group that can be further divided into aquatic and semiaquatic mammals.Aquatic mammals are highly adapted and confined to water for all aspects of their life and cannot travel on land, such as the Amazon River dolphin (Inia geoffrensis) (Pacini & Harper, 2008;Veron et al., 2007).Semiaquatic mammals are less adapted to aquatic life (Dunestone & Gorman, 1998;Howell, 1930).They require land for nesting, rest, or reproduction but display an obligate connection with water in their daily activity, especially as a main food source (Hood, 2020).This includes terrestrial mammals that require water sources for food and protection but excludes species reliant only on freshwaterassociated habitats, such as surrounding grasslands and not freshwater systems per se.For example, the webbed-footed tenrec (Limnogale mergulus), which derives its food from the water, and swamp rabbit (Sylvilagus aquaticus), which utilizes waterbodies to avoid predation, are semiaquatic.However, jaguars (Panthera onca), which hunt predominantly around waterways, and puku (Kobus vardonii), which graze in grassland surrounding waterways, are excluded.
Due to the global scale of the threats to freshwater ecosystems, there is an urgent need to quantify life-history traits associated with higher extinction rates in freshwater species.The International Union for the Conservation of Nature (IUCN) Red List is used to document the world's changing biodiversity and identify species most at risk of extinction and in need of conservation (IUCN, 2021).Combining IUCN assessments with data on life-history traits within taxonomic groups provides insight into why some species are more susceptible to extinction and can identify potentially threatened species that have yet to be evaluated or lack sufficient data for IUCN Red List classification (Hernández-Yáñez et al., 2022;Kopf et al., 2017;Marneweck et al., 2021).In this review, we aimed to document the key threats to freshwater mammals based on information in the IUCN Red List.We examined species' taxonomy, biogeographic location, habitat characteristics, and life-history traits to determine species' associations with extinction risk and identify groups of the greatest conservation concern.

Species list
We compiled a comprehensive list of freshwater mammals from multiple sources.All mammalian species listed in version 2021-3 of the IUCN Red List (IUCN, 2021) as occupying freshwater systems were included (n = 146) as were additional species on the Freshwater Animal Diversity Assessment (FADA) (Balian et al., 2008;Véron et al., 2010), which includes aquatic and water-dependent species (n = 152).Because the definition of semiaquatic mammals varies greatly and new species have been described since 2010, we also consulted Hood (2020) on semiaquatic mammals (n = 140).These 3 sources were combined to produce a list of 203 species.
To ensure all species conformed to the scope of our review, we excluded volant species, species that do not occupy freshwater habitats, and riparian species that only use the surrounding vegetation for food or hunting grounds.We used a precautionary approach to include species deemed reliant on freshwater environments when basic ecological information was unknown.Due to the differences in levels of aquatic specialism of semiaquatic mammals, we ranked these species on 3 levels: high, moderate, and low (Appendix S1).Species with a high aquatic affinity can dive, remain submerged for extended periods, and display numerous physiological or morphological aquatic adaptations (e.g., webbing, natatory fringes, or ability to close all orifices) (Howell, 1930).Moderate species are less biologically specialized but still display some level of aquatic adaptation or behavior and instinctively seek water for protection and main dietary items (Dunstone & Gorman, 1998).Lowlevel species are highly associated with water, and likely seek water to avoid predation, but exact interactions are either unknown (such as rare or data deficient species) or limited to the shallows of the water's edge.Fully aquatic mammals were classified as such for comparison with semiaquatic mammals.

IUCN Red List categories and criteria
The IUCN (2012a) Red List acknowledges 9 categories of extinction risk: extinct (EX), extinct in the wild (EW), critically endangered (CR), endangered (EN), vulnerable (VU), near threatened (NT), least concern (LC), data deficient (DD), and not evaluated (NE).Threatened categories (CR, EN, and VU) are assigned based on 5 criteria that describe different aspects of extinction risk (IUCN, 2012a): criterion A, population reduction; B, small, fragmented, or fluctuating geographic range; C, small and declining population size; D, very small or restricted populations; and E, high probability of extinction assessed through quantitative analyses.
We recorded the current IUCN Red List category for each species and criteria for inclusion in our review as of April 2021.Because uncertainty about the true extinction risk of species is introduced by DD and NE statuses, the percentage of threatened species was estimated at the best level and its lower and upper bounds (IUCN, 2021).The best estimate of extinction risk was based on the assumption that the proportion of DD and NE species follows that of threatened data-sufficient species, that is, (CR + EN + VU) / (N -EW -EX -DD -NE), where N is the total number of species included in our review.The numerator indicates the number of threatened species in the scenario, and the denominator is the total species being considered.The lower estimate is based on the assumption that all DD and NE species are not threatened: (CR + EN + VU) / (N -EW -EX).The upper estimate is based on the assumption that all DD and NE species are threatened: (CR + EN + VU + DD + NE) / (N -EW -EX).Population trend was recorded as either increasing, stable, decreasing, or unknown for all extant species.

Red List Index
The scale for the Red List Index (RLI) ranges from all taxa are EX (0) to all taxa are LC (1) (Butchart et al., 2007).The RLI tracks trends in the extinction risk of a group of species over time and includes only genuine reasons for species category changes.New information, changes in red-list criteria, and taxonomic revisions are not genuine reasons.We used backcasting to correct previous categories associated with changes that were not genuine and excluded species already EX or EW (Bubb et al., 2009).To calculate the RLI (Equation 1), we multiplied the number of evaluated non-DD species in each category by the assigned category weight (LC = 0; NT = 1; VU = 2; EN = 3; CR = 4; CR [Possibly Extinct], EX and EW = 5) (Butchart et al., 2004;IUCN Standards & Petitions Committee, 2022).The sum of these products was then divided by the total number of evaluated and data-sufficient species extant at the first year multiplied by 5.This number was subtracted from 1 to produce an RLI for the particular year the categories were taken.
Due to different time delays between species assessments, RLI values were calculated only for 1996, 2008, and 2016.

Taxonomic and geographic patterns of extinction risk
We assessed taxonomic patterns of extinction risk based on the proportion of threatened species in each order and family.We recorded the biogeographic location of each species based on ranges reported in the IUCN Red List or in the literature for NE species.Location was categorized into 6 of the 8 biogeographical realms where freshwater mammals were found: Afrotropical, Australasian, Indo-Malayan, Nearctic, Neotropical, and Palearctic (Olson et al., 2001).We mapped the species richness for available species and threatened species on a global scale from shape files obtained from the IUCN Red List mammals data set (IUCN, 2021).We read relevant files into R Studio (R Core Team, 2022), with the package rgdal (Bivand et al., 2022), and retrieved species ranges with the unique function of the raster package (Hijmans, 2022).The unique layers were converted to a raster object and plotted onto world maps with the tmap package (Tennekes, 2018).
We obtained habitat data from the IUCN Red List, or relevant literature, and included subcategories of wetland habitats (IUCN, 2012b).Habitat breadth was recorded as the number of habitat areas the species could reside in and wetland breadth the number of wetland types the species occupied.We assigned a broad climate type (tropical, arid, temperate, cold, or polar) to each species based on their range overlap with the Köppen-Geiger climate map (Peel et al., 2007;Appendix S2).

Threatening processes
We recorded direct threatening processes listed in the IUCN Red List for all evaluated species.Threats were classified into 11 groups according to version 3.2 of the IUCN (2012c) Threats Classification Scheme and included residential and commercial development, agriculture and aquaculture, energy production and mining, transport and service corridors, biological resource use, human intrusions and disturbance, natural systems modifications, invasive and other problematic species, genes and diseases, pollution, geological events, and climate change and severe weather (IUCN, 2012c).Some species were recorded as not currently facing any species-level threatening processes and facing only minor threats on a subpopulation level or their current threats were unknown.Threats were recorded for ongoing and expected future threatening processes.Species affected by the top 6 threatening processes were mapped to identify where they are most at risk.

Life-history traits and extinction risk
We recorded life-history traits from the COMBINE (a Coalesced Mammal database of Intrinsic and Extrinsic traits) data set (Soria et al., 2021), IUCN Red List (IUCN, 2021), or relevant literature.Traits included body mass, trophic level, dietary breadth (amount of plant or animal groups or both that made >20% of the diet), activity period (nocturnal, diurnal, or mixed [i.e., cathemeral or crepuscular]), dispersal distance, and home range size.Sexual maturity, life span, and generation length were highly correlated (r > 0.85), so only generation length, with the largest sample size, was retained.Additional predictors of extinction risk included the level of aquatic specialization, realm, climate, habitat breadth, and number of threats a species is currently facing.Because range size is used in assessment criteria B for threatened species, it was not considered for analysis.
We used multilevel models to determine whether extrinsic factors and life-history traits contributed to extinction risk in freshwater mammals.The binomial threat status of threatened (EX, EW, CR, EN, and VU) and nonthreatened species (NT and LC) was used as the response variable for all models.Unevaluated and DD species were not included, leaving 137 species to model.We fit Bayesian Bernoulli univariate regression models with the brms package (Buerkner, 2017) in R Studio (R Core Team, 2022).Each model contained 2 variables: a fixed-effect variable and a random-effect variable, controlling for taxonomic differences (order and family).We ran models with 4 chains of 2000 iterations with the first 1000 iterations used as a warmup for a total of 4000 postwarmup draws.A control value of 0.99 was fitted to prevent divergent transitions; convergence was achieved for all models (Rhat = 1.00).We calculated the probability of direction (pd) to determine the existence of an effect by each parameter, which was deemed significant when the 95% credible interval excluded zero.Plots were made with the ggplot2 and ggeffects packages (Lüdecke, 2018;Wickham, 2016).

Categories and criteria
A total of 166 species remained after all inclusion criteria were met, representing 10 orders and 30 families (Appendix S1).Of these species, 157 have been evaluated by the IUCN.Threatened species comprised 30.1% of freshwater mammals, with 21 species (12.7%) listed as VU, 26 species (15.7%) listed as EN, and 3 species (1.8%) listed as CR.The Ethiopian amphibious rat (Nilopegamys plumbeus) and Baiji (Lipotes vexillifer) were flagged as CR possibly extinct.Three species (1.8%) were EX and 1 was EW (Appendix S1).Twenty species (12.0%) were listed as DD, 13 of which had previously been categorized but now lack sufficient information regarding taxonomy, population status, or range size (Appendix S3).Sixty-nine species (41.6%) were listed as LC and 14 species (8.4%) as NT.
Most threatened species (49.2%) were classified under criterion A (subcriteria A2, A3, and A4) due to large population reductions.Small, fragmented, or fluctuating geographic range size characterized 28.8% of threatened species, predominantly based on the extent of occurrence (B1) or total spread of the species' range (known for 107 species), rather than the area of occupancy (B2, known for 18 species).Small population size and decline, where the number of mature individuals is typically <10,000, accounted for 18.6% of threatened species (criterion C).Only the Venezuelan fish-eating rat (Neusticomys venezuelae) and Baiji were classified under criterion D due to their very small, restricted population containing <1000 mature individuals.No freshwater mammals were categorized by predictive quantitative analyses under criterion E.
The best estimated proportion of freshwater mammals that are threatened, based on the current proportion of evaluated threatened species, was 37.6%.The final range, from all unevaluated species being nonthreatened to all being threatened, was 30.9-48.8%.Of the 50 threatened species, 44 species faced decreasing population size, and the population trends of 4 species were unknown.Only the EN Indus river dolphin (Platanista minor) had an increasing population, and VU hippopotamus (Hippopotamus amphibius) had a stable population (Appendix S4).Overall, 27 species had stable populations, and the Eurasian beaver (Castor fiber) was the only other species with an increasing population trend.Unknown population trends were listed for 60 species, and decreasing trends were observed for 75 species.

Red List Index
The RLI showed temporal change in the extinction risk of species assemblages.The RLI was calculated for 132 species that had been evaluated based on IUCN Red List criteria and not reported as DD or EX during the first assessment following backcasting.Not all species were reported on in the same year; new publications and category changes occurred periodically.Because the RLI requires the same number of species present each year, and many species had gaps of up to 15 years between assessment publications, we used only 3 reporting years: 1996, 2008, and 2016.Six of the 7 genuine changes that occurred after 2016 were backcast to that year because all assessments took place either in 2016 or early 2017, and the driver of the genuine change was already present.The last genuine change occurred in 2021 where the Pyrenean desman (Galemys pyrenaicus) was upgraded to EN and was not included in the calculation.The RLI for freshwater mammals (Appendix S5) in the 12 years between 1996 and 2008 showed a decline of 2.0% (compared with 0.8% decline in the global mammal assemblage [Hoffmann et al., 2011]) and over 20 years a decline of 3.6% (compared with a 1.6% decline in all freshwater species [IUCN, 2021]).

Geographic patterns of extinction risk
Globally, freshwater mammals were widely distributed (Figure 1a), with most species occurring in the Neotropical (56) and Afrotropical (48) realms, followed by the Palearctic (22), Indo-Malayan (17), Nearctic (13), and Australasian (10) realms.Ninety-seven species occurred in tropical climates compared with 45 species in temperate areas, 17 in cold climates, and 7 in arid regions.No freshwater mammals occurred prominently in polar climates, despite some species' extent of occurrence overlapping with small polar zones.
The Neotropical realm supported 15 threatened species (26.8%) and the Indo-Malayan realm 13 threatened species (76.5%).Eleven threatened species occurred in the Afrotropical realm, 10 occurred in the Palearctic realm, and only the water mouse (Xeromys myoides) was threatened in the Australasian realm (Figure 1b).Conversely, all species in the Nearctic realm are considered LC, with the exception of the DD Glacier Bay water shrew (Sorex alaskanus).Of the 29 unclassified DD and NE species, 13 were from the Neotropical realm, 9 from the Afrotropical, 4 Australasian, 2 Indo-Malayan, and 1 Nearctic.

Threatening processes
All threatening processes listed by the IUCN affected at least 1 freshwater mammal at 1 point in time.However, currently no geological events were considered a species-level threat.The most prevalent ongoing threats included biological resource use, agriculture and aquaculture, pollution, residential and commercial development, and modification of natural systems (Figures 2 & 3).Cropping practices, hunting, logging, agricultural runoff, housing development, and water management practices were the leading threats within the main threat categories.Transport corridors, particularly roads, railway, and shipping lanes, threatened 25 species, and current threats of 16 evaluated species were unknown.Invasive, problematic species, genes, and diseases, transport corridors, mining and energy production, and human intrusion of natural habitat each affected <12% of species (Figure 2).Minor threats that did not affect the entire population of a species, such as hunting for food or pelts, habitat loss and degradation, or poisoning of species considered pests, were reported for 15 species.Climate change effects were reported for 19 species, represented by drought (13 species), habitat shifting and alteration (7 species), storms and flooding (6 species), and temperature extremes (2 species).Of those species facing known threats, the majority were threatened by 3 (20 species) or 4 (18 species) separate processes, with a maximum of 9 effecting any 1 animal (Appendix S9).
More than half of all ongoing threatening processes were reported in the Neotropical (27.6% of threats) and Afrotropical (24.0%) realms.The Palearctic followed (19.8%), then Indo-Malayan (16.4%),Nearctic (7.6%), and Australasian (4.5%) realms.The Afrotropical realm contained the highest number of species threatened by biological resource use (20 species), agriculture and aquaculture (19 species), human disturbance (5 species), and unknown factors (14 species) (Figure 3).Pollution (19 species), agriculture and aquaculture (19 species), residential development (12 species), energy production (7 species), and minor threats (5 species) were the most prevalent in the Neotropical realm.However, this realm also contained the greatest number of species not experiencing current threats (10 species).The Palearctic realm contained the greatest number of species threatened by ecosystem modification (12 species), invasive species and diseases (7 species), transport corridors (7 species), and climate change (7 species) and showed a high concentration of species affected by water pollution (13 species).The Indo-Malayan realm contained a high proportion of species threatened by resource use (16 species), agriculture (15 species), and urban and residential development (11 species).

Species richness
The IUCN assessments of 18 species predicted they will be affected by further threatening processes in the next century, including climate change, housing development, mining and wood plantations, dams and surface water abstraction, industrial and military effluents, and noise pollution.This included 8 species of otters-3 of which are currently threatened-that will be affected by habitat alteration, drought, and flooding.The 2 extinct Madagascan hippopotamuses were likely threatened by overhunting (Hansford et al., 2021), and Nelson's rice rat (Oryzomys nelsoni) was likely driven to extinction by competition with the invasive black rat (Rattus rattus) (Timm et al., 2017).The extinction of Père David's deer (Elaphurus davidianus) in the wild was attributed to habitat loss associated with housing development, agricultural crops, and hunting (Jiang & Harris, 2016).

Life-history traits and predictors of extinction risk
Despite the majority of freshwater mammals being listed as LC, many species remain poorly known and basic data on lifehistory traits are lacking.For instance, home range data were only available for 34 species.Body mass was a strong indicator of extinction risk; the greater the mass, the more likely to be threatened (Table 1; Figure 4).We also found the probability of a species being threatened increased as dispersal distance and number of threats to a species increased (Table 1; Figure 4).Habitat breadth, wetland breadth, activity cycles (diurnal, nocturnal, or mixed), and trophic level had no effect on extinction risk (95% CI contained zero).Although the degree of aquatic specialization in semiaquatic mammals had no effect on their risk of extinction, fully aquatic mammals were at a significantly greater risk of extinction than semiaquatic mammals (Table 1; Figure 5a).

DISCUSSION
We considered the extinction risk of 166 species of freshwater mammals in relation to their current and historical conservation status in the IUCN Red List.Despite gaps in knowledge of the basic biology of many species, particularly those listed as LC, it was clear that freshwater mammals had a higher risk of extinction and rate of decline compared with the global assemblage of mammals and freshwater animals in general.The best estimate of the true proportion of threatened species was 37.6%, which is higher than all freshwater taxa (32%; Collen et al., 2014) and terrestrial mammals (26%; IUCN, 2021).Similarly, freshwater mammals had a much greater proportion of species with declining populations (44.0%) than freshwater vertebrates (26.3%) and all mammals (32.0%) (IUCN, 2021;WWF, 2020).The rate of decline of 3.6% over 20 years equates to the IUCN Red List status of 6 species increasing by at least 1 category within the next 2 decades.With the world's vertebrate populations having already declined 68% in the past 50 years, this is a concerning trend for freshwater mammals (WWF, 2020).

Dangers of data deficiency
The elevated extinction risk of freshwater mammals highlights the need for greater study of DD and unevaluated species.Datadeficient species exhibit a higher probability of being threatened than data-sufficient species, making the status of DD particularly precarious by masking the severity of their true situation (Bland et al., 2015;Borgelt et al., 2022;Dudgeon, 2020).Predictions of the extinction risk of DD mammals are that 63.5% are already threatened (Bland et al., 2015), suggesting that the pessimistic upper estimate of the proportion of threatened freshwater mammals of 48.8% may be more accurate.A lack of current population assessments is responsible for the lapse in categorization of many DD species, which highlights the importance of developing monitoring programs.Here, as much as 65% of DD freshwater mammals had been along with other highly threatened species, such as the EN Amazon River dolphin geoffrensis) (da Silva et al., 2018), as DD for assessments before being uplisted from their previous status.
Most DD freshwater mammals small rodents or shrews (<100 g), which either inhabit areas that are difficult to survey (e.g., Rowe et al., 2014) or are taxonomically uncertain, such as Dasymys spp.and potentially Colomys goslingi (Giarla et al., 2021;Mullin et al., 2005).Resolutions of species complexes are important, for instance, 2 Dasymys species that contained 6 newly described species were previously considered LC based on their large geographic range (Mullin et al., 2005;Taylor, 2016).These ranges are now known to be disjunct, with the potential for higher threat classification, for example, the EN montane shaggy rat (Dasymys montanus) with its fragmented and declining range (Kennerley, 2016).Such poorly known and newly erected or described DD mammals are among the most likely to be threatened (Bland et al., 2015;Liu et al., 2022;Padial & De la Riva, 2006).Therefore, the IUCN Red List must keep up with increasing taxonomic inflation (Padial & De la Riva, 2006).
Other methods of assessment should also be considered when evaluating species that may lack sufficient population or geographical range data.Although no freshwater mammals were categorized under criterion E (quantitative analysis), using remote methods such as population viability analysis or estimating extinction risks from genetic or habitat status may help prevent species being listed as DD (Mace et al., 2008;Wilder et al., 2023).The use of historic range data in determining the degree of range reduction under criterion B may present another option for difficult-to-survey species, such as the platypus (Ornithorhynchus anatinus), which has had dramatic range declines, highlighting its need for reassessment (Hawke et al., 2019).

Threatened species hotspots
Tropical areas have the richest biological diversity of vertebrates (Ceballos & Brown, 1995), followed by temperate regions (Mace et al., 2005), and this pattern holds true for freshwater mammals.Our species richness maps revealed the Neotropical and Afrotropical realms supported the highest concentrations of species, with the highest proportion of threatened species residing in the Indo-Malay-particularly in Malaysia and Indonesian islands of Sumatra and Kalimantan (Figure 1).This region has the highest rate of deforestation globally, driven by logging, crop plantations (particularly palm oil), road construction, and residential development (Dudgeon, 2022;Hughes, 2018;Stibig et al., 2014).Current measures to reduce deforestation have been unsuccessful; clearing continues to increase with demand for housing and plantation products, despite this process being identified as a major threat to biodiversity in the region (Hughes, 2018).Population declines of freshwater megafauna are also greatest in the Indo-Malay region, further attributed to overexploitation and dam construction (He et al., 2019).
Other hotspots of threatened species included the Palearctic (particularly France, Spain, and Russia), Neotropics (Brazil, Venezuela, Bolivia, and Paraguay), and fragments of the Afrotropics (South Sudan).The main threats identified included biological resource use, agriculture, residential development, pollution, and ecosystem modification, where the greatest effects were evident in the Palearctic.The most prevalent impact of ecosystem modification was from water management practices and dams.Although the importance of biogeographical region was clear, the number of occupied habitats did not significantly affect species extinction risk, despite narrow habitat breadth (indicating a more specialized habitat niche) being one of the strongest traits indicative of species more prone to extinction (Chichorro et al., 2019).It is uncertain why this occurred, although it is likely associated with the high levels of total human disturbance (Riggio et al., 2020) in realms with high proportions of threatened freshwater mammals.Significant relationships between habitat type and extinction risk may be observed if species are assessed on a finer scale (e.g., subspecies or populations).

Hunting, logging, and agriculture
Overall, the greatest threats to freshwater mammals are biological resource use-through hunting and logging-and agriculture.Targeted hunting and overexploitation affected almost one third of freshwater mammals, including 21 protected species.Overharvesting of freshwater resources is largely associated with the ease of accessibility and exploitation of freshwater areas compared with hunting in expansive terrestrial refuges (Antunes et al., 2016;Dudgeon, 2020).Freshwater aquatic species are also less resilient to overexploitation; the effects of historic hunting practices still affect current populations (Antunes et al., 2016).Agricultural practices affect freshwater mammals directly via habitat loss as well as indirectly by contributing to other threatening processes (Erisman et al., 2016).Within the IUCN agricultural threat categories, cropping and livestock production were the greatest threats.Indirectly, agricultural effluents contributed to over one third of the species threated by pollution.Agriculture is also the largest consumer of freshwater, contributing to the effects of water abstraction (Shiklomanov, 1993;Stoett et al., 2019).
Interestingly, two thirds of the threats freshwater mammals face are based in the terrestrial system, largely due to crops and livestock, expanding housing developments, timber plantations, mining or road development, and the consequential land clearing often required.Similarly, the effects of habitat loss and degradation-particularly from logging and urban development-affect >80% of all freshwater animals (Collen et al., 2014).Over one quarter of threats potentially affect both terrestrial and aquatic habitats, including pollution, human recreational activities, ecosystem modification, drought, or introduced species.Subsequently, only 12% of threats affect the aquatic system directly, including aquaculture, shipping lanes, fishing, harvesting of aquatic resources, and, most predominantly, dams and water management.Water management practices currently threaten 37 freshwater mammals with many overarching effects.Large impoundments and altered downstream hydrology can cause habitat fragmentation, drive in availability, and restrict movement across the landscape (Barbarossa et al., 2020;Singh et al., 2021;Soukhaphon et al., 2021;Wu et 2019).Particularly for fully aquatic mammals, dams are a barrier for dispersal that can drastically restrict gene flow, as has been seen with the Irrawaddy dolphin (Orcaella brevirostris), which persists in 3 declining populations along the Mekong River (Dudgeon, 2022).Altered flows and cold-water releases further affect foraging, nesting, or breeding behaviors of several freshwater mammals (Escoda et al., 2019;Hawke et al., 2019;Pedroso et al., 2014;Wu et al., 2019).Dams and water traffic also contribute to noise pollution, which can increase stress and affect echolocation and communication efficiency of aquatic mammals (Dey et al., 2019).

Predisposition for extinction
Life-history traits have been used to assess the likelihood of species extinction or decline and predict IUCN Red List categories of unevaluated species (Bland et al., 2015;Cardillo, 2003;Hoffmann et al., 2011;Kopf et al., 2017;Rija et al., 2020).Here, only body mass and dispersal distance had a significant influence on the extinction risk of species.In a trend following that for all mammals and freshwater megafauna, heavier freshwater mammals are at a higher risk of extinction (He et al., 2019;Hernández-Yáñez et al., 2022;Hoffmann et al., 2011;Rija et al., 2020;Ripple et al., 2017) because they typically have longer life spans and generation lengths, greater maturation ages, longer gestation periods, and lower reproductive output (Cardillo, 2003;Cardillo et al., 2005;He et al., 2019;Stearns, 1983;Whitmee & Orme, 2013).Species that disperse long distances are exposed to a greater number of potential threats when moving across large areas, which makes them susceptible to habitat alteration and increases their risk of extinction.This is a common issue for potamodromous fish, which migrate purely within freshwater systems and are highly threatened by reduced connectivity (Deinet et al., 2020;Dudgeon, 2020).
Although large freshwater mammals with greater dispersal distances are of greatest conservation concern, small species, which represent the majority (107 species <1 kg), should not be ignored.Small species receive less attention and research than large species, despite many being at the same risk of endangerment (Bland et al., 2015;Marneweck et al., 2021).Rodentia was the largest order represented in our analyses, yet even poorly known and threatened rodents were not recommended for further research in their IUCN assessments (Rivas, 2018;Weksler & Timm, 2017).Indeed, it is highly likely that there are species, such as the rakali (Hydromys chrysogaster), that require an elevated risk status but remain LC owing to a lack of basic ecological knowledge (Williams, 2019).

Recommendations for freshwater mammal conservation
Our results imply that the number of threats freshwater mammals face can combine to increase their risk of extinction.With the increasing prevalence of threats to freshwater environments and interactions emerging between threatening processes (Geary et al., 2019), it is important efforts are taken to manage impacts from multiple stressors (Reid et al., 2019).Given the threats we identified, freshwater mammals would benefit from improved protection of freshwater and surrounding riparian habitats; greater hunting prevention; improved management of industrial, domestic, and agricultural pollution; and greater freshwater connectivity.
Greater protection and restoration of freshwater systems and surrounding habitats may mitigate the stresses induced by agriculture, urban and industrial development, and logging.Protection can be provided through formal means, such as with protected areas (listed locally or through initiatives such as the Ramsar Convention) or through policy aimed to maintain freshwater connectivity (Maasri et al., 2021;Tickner et al., 2020).The installation of buffer and livestock exclusion zones around freshwater habitats can also be achieved on smaller scales by landholders (Kauffman et al., 2022;Luke et al., 2019;McCormack, 2023;Rhodes et al., 2002;Singh et al., 2021).Such methods not only improve general habitat quality, but also act as a filter to reduce nutrient runoff.Although previous legislative efforts have been unsuccessful in curbing freshwater pollution (Dudgeon, 2020), improved wastewater treatment, regulation of industrial pollution, and sustainable agricultural practices have led to promising improvements in water quality (McCormack, 2023;Tickner et al., 2020).Improved fishing regulations, antipoaching strategies, and poverty alleviation to limit overharvesting, bycatch, and illegal hunting can also mitigate threats associated with biological resource use (Dudgeon, 2020;Rija et al., 2020;Tickner et al., 2020).Lack of water connectivity created through intense regulation of waterways and dams is a significant threat to freshwater biodiversity.We recommend the ecological and hydrological needs of freshwater mammals be considered in all future water infrastructure plans.Environmental flows ("the quantity, timing, and quality of freshwater flows and levels necessary to sustain aquatic ecosystems" [Arthington et al., 2018, p. 4]) are another key to restoring freshwater ecosystem function (Dudgeon, 2022;McCormack, 2023).Further methods to restore natural watering and flooding processes include the removal of obsolete dams and levees and greater protection of remaining free-flowing systems to prevent negative effects of river regulation (Dudgeon, 2020;He et al., 2021;Tickner et al., 2020).

ACKNOWLEDGMENTS
The IUCN is acknowledged for the use of its information, range maps, and classification of species.

FIGURE 1
FIGURE 1 Species richness of (a) all freshwater mammals with available range maps (n = 153) and (b) threatened freshwater mammals (n = 50, including extinct in the wild).

FIGURE 2
FIGURE 2 Number of freshwater mammals affected by each threatening process and the proportion (pie charts) of threatened species affected based on the International Union for the Conservation of Nature Threats Classification Scheme 3.2 (IUCN, 2012c) (first 5 circles on the left, 5 greatest threats; threats aligned to the right of top 5 threats, respective subcategory threats; bottom row, other threats that affect freshwater mammals).

FIGURE 4 FIGURE 5
FIGURE 4The effect of (a) body mass (n = 124 species) (limited to 350 kg because the proportion of threatened species plateaus when body mass is >300 kg [max body mass 1500 kg]), (b) compounding threatening processes (n = 117), and (c) dispersal distances (n = 125) on the threatened status of freshwater mammals evaluated by the International Union for the Conservation of Nature (shading, 95% credible interval).
Funding was provided by the Australian Government Research and Training program, Gulbali Accelerated Publication Scheme, and the Murrumbidgee Monitoring Evaluation and Research program funded by the Commonwealth Environmental Water Office.Open access publishing facilitated by Charles Sturt University, as part of the Wiley -Charles Sturt University agreement the of Australian University Librarians.ORCID Emmalie Sanders https://orcid.org/0000-0002-1051-1854

TABLE 1
Results of univariate Bayesian regression models predicting the effect of life-history and extrinsic factors on the extinction risk of freshwater mammals.
a Probability of direction, which describes certainty of the effect direction.b Credible intervals in parentheses.c Significant result at the 95% credible interval.d Comparison of extinction risk between aquatic mammals and semiaquatic mammals.