1. Introduction
Freshwater aquatic ecosystems are under increasing threat from human activity and climate change. They are among the most heavily altered ecosystems with disproportionately high biodiversity loss worldwide [
1,
2]. Major systematic drivers of aquatic species loss include land cover and land use change (LUCC), overexploitation, invasive species, and climate change [
3,
4,
5,
6,
7,
8]. Direct and indirect competition for water resources with humans (e.g., water abstraction for irrigation) impose stress upon aquatic ecosystems and imperil fauna [
9]. Aquatic ecosystems are particularly vulnerable because they receive the cumulative effects of stressors within the watershed. Reconstruction of a region’s LUCC history can be used to determine the potential stressors that have impacted or degraded habitats over time and thereby can enhance current and future management [
10]. While some sources of aquatic habitat degradation such as the construction of manmade dams and river impoundments are obvious, others are more obscure. For example, land burning as the primary cause of eutrophication in Lake Victoria was only recently determined [
11].
Combining standardized geospatial data of population and climate (e.g., [
12,
13,
14]) with remotely sensed data can produce spatially explicit models of a region’s environmental geography, distribution of endangered species and biodiversity at regional and global scales [
15]. Satellite imagery provides critical information for assessing changes in terrestrial ecosystems that can be linked to effects on aquatic ecosystems [
16]. In Tanzania, an assessment of the Usangu Plains wetlands and Malagarasi River drainage concluded that these large river catchments suffered substantial land cover changes with declines in woodland and wetland habitats co-occurring with increases in settlements, agriculture and grassland [
17,
18]. In the Usangu Plains wetlands in particular, vegetated swamp cover decreased by 67% between 1984 and 2001 [
18].
Relatively little is known about the endemic African fishes outside of the Great Lakes (primarily Lakes Victoria, Tanganyika and Malawi). Habitat specializations, phylogeny, reproduction and diet have been sparsely documented for fishes inhabiting the smaller lakes, wetlands and river systems in Tanzania, while LUCC and climate change effects on the species are almost entirely unknown [
19]. Many endemic fishes are highly specialized to their local environment (e.g., rheophilic cichlids and thermal-alkaline-salt tolerant cichlids). The survival of others, such as annual killifish that rely on ephemeral pools in the savannah, is critically dependent upon the land use and land cover [
20]. Even recently, new species of fishes have been discovered in areas affected by LUCC, for example, three new species of sucker-mouth catfishes (genus
Chiloglanis) were found in the lower Malagrasi River [
21], and a previously unknown rheophilic cichlid,
Haplochromis vanheusdeni (Schedel, Friel & Schliewen 2014), was discovered in the Great Ruaha River drainage [
22]. Others such as the characin
Petersius conserialis (Hilgendorf 1894), from the Ruvu and Rufiji Rivers are only known from past sightings. LUCC also affects endemic species in lakes, such as the cichlids inhabiting African crater lakes (e.g., Lakes Ejagham, Bermin, Barombi-Mbo, and Bosumtwi) that evolved in small bodies of water and are sensitive to environmental change. Some of these cichlids are also highly specialized in their diets, for example
Pungu maclareni (Trewavas 1962), and
Coptodon spongotroktis (Stiassny, Schliewen & Dominey 1992), feed predominantly on freshwater sponges [
23]. The primary threats to the aquatic fauna of these crater lakes are siltation due to LUCC and water extraction for agriculture and other uses [
24].
In the context of this study, we adopt the definition of vulnerability provided by the fourth assessment report from the Intergovernmental Panel on Climate Change (IPCC) as “the degree to which a system is susceptible to, and unable to cope with, adverse effects of climate change, including climate variability and extremes” [
25]. Climate change projected for East Africa, including decreased precipitation and increased temperature affect freshwater fish vulnerability by lessening the capacity of water bodies to regulate against temperature change, altering the thermal suitability of the habitats [
26] and, in extreme cases, the spatial extent of the habitats and resulting in a decrease in the natural ecosystem’s ability to mitigate against the effects of climate change. Native riparian vegetation is known to lessen the impacts of LUCC and climate change on aquatic ecosystems by improving water quality [
27,
28,
29,
30], moderating water temperature [
29,
31], providing food and resources [
32] and improving biodiversity [
29,
31,
33]. Therefore, degradation or conversion of the riparian vegetation to other land cover types such as crops unavoidably results in the deterioration of aquatic ecosystems [
34]. In addition to the integrity of the vegetation adjacent to aquatic habitats, historical LUCC (i.e., land use legacy) at larger landscape scales has been shown to be important for fish and invertebrate diversity [
10,
35]. Aquatic habitat quality is dependent upon the extent of LUCC at catchment or regional scales [
10]. Furthermore, the rarity of many individual species within diverse freshwater communities often hinders conservation efforts. A lack of thorough natural history data on rare species results in extreme difficulties in forecasting the identity of taxa affected by climate change or LUCC at large scales [
24]. There is a fundamental gap in knowledge of the vulnerability of endemic species to LUCC, climate change and socio-economic factors that threaten their habitats.
The overall objective of this study was to examine the LUCC from satellite imagery over the 1984–2015 period for three freshwater habitats of endemic cichlids. We also summarize putative climate and landscape stressors including population growth and forecasted climate change.
The family Cichlidae contains 1693 described species (as of May 2017), with many new species yet to be described. Cichlids are distributed exclusively in tropical fresh water, have colonized most major bodies of water, and share these habitats with a great diversity of other aquatic species. Because many have limited ranges and high levels of specialization and/or endemism, they are good indicators for overall aquatic habitat degradation from LUCC and climate change.
4. Discussion
The LUCC history around aquatic habitats has been shown to have an important influence on the present-day fish diversity and habitat conditions [
10,
78]. However, the classification of satellite imagery in the arid ecosystems examined here is inherently challenging. In an assessment of LUCC for the slopes of Mt. Kilimanjaro, it was indicated that from single date imagery, bushland, grassland and cropland cannot be readily separated due to the transitional nature of the classes and their spectral similarity [
48]. Furthermore, vegetation dynamics in the region are strongly influenced by cultivation and land conversion [
16]. Small-scale agriculture, with variable irrigation practices can be difficult to separate from the natural vegetation phenological cycles in arid regions [
79]. However, separating the natural landscape from modified areas is needed in order to understand trends in LUCC dynamics [
69]. The tasseled cap time sequence of the historical Landsat imagery allowed us to separate out the natural land cover classes from the modified areas and reduce the confusion between spectrally similar classes at single points in time. The results also illustrate an isolation and fragmentation of the natural vegetation around Lake Chala and Chemka Springs. In both cases the total native vegetation area was found to be below the recommended 70% for the maintenance of water quality [
30]. The width of the Riparian buffer around Chemka springs is below the recommended thresholds for aquatic wildlife protection (30 m), and is at the lower limit of the thresholds for stream temperature stabilization (10 m), nutrient retention (15 m), sediment control (10 m), and pesticide retention (15 m) [
80]. However, when land use intensity is considered, the minimum buffer recommended to moderate water temperature and quality increases to 30 m [
81]. In contrast, for Lake Chala the vegetated slopes of the crater and the bushland vegetation on the Tanzanian side afford some protection but this is greatly reduced on the Kenyan side of the lake where, outside of the crater the natural vegetation has been removed. If the rate of natural vegetation loss were to remain constant, the remaining fragments of bushland in the Lake Chala catchment would be converted to other land covers by 2040 and the remaining closed canopy forest around Chemka Springs would be lost by 2033. Mitigation and management efforts could in the long-term to protect the remaining natural vegetation and potentially also to allow it to regrow.
The rapid decline of groundwater in the region followed by the increase in water availability can be seen in the mascon grid time series for our study areas (
Figure 9A,B). With the rainfall patterns (and associated vegetation phenology) in East Africa closely associated with IOD and El Nino Southern Oscillation (ENSO) events, further depletion events in terrestrial water storage can be expected [
16,
82]. Depletion events would be important for small aquatic habitats where subsurface sources (e.g., springs) are the primary source of incoming water. Irrigation and other water use such as drinking water in the region almost entirely draws water from sources originating from groundwater [
42,
47]. With precipitation being a significant driver of the water storage behavior of the region encompassing our study areas [
83], climate change scenarios for the region suggest an increase in water stress in the future. In addition, interannual high precipitation events associated with large-scale circulations (i.e., IOD, ENSO) will play increasingly important roles in mitigating accumulated water storage deficits [
83]. A global assessment by Reference [
84] found high correlations between threats to human water security and biodiversity, with the strongest correlations in heavily settled regions indicating the negative influence a growing population has on aquatic ecosystems.
The continent-wide approach by Reference [
19] assessed ecoregions based on land degradation and threats to aquatic habitats. Our approach focuses on a finer scale by considering current and future stressors to the specific aquatic habitats of cichlid species taking into account the global and regional drivers of ecosystem vulnerability recommended by Reference [
85]. Of the habitats examined here, we found that the
Alcolapia species flock restricted to the Lake Natron Springs was vulnerable to forecasted climate change and potential future landscape alteration. The water in the lake itself is uninhabitable for fish to survive, even a short period of time. Therefore, if the temperature of the shallow springs surpasses their thermal tolerance or the water conditions are altered by future water abstraction [
86,
87], these fish have no migratory recourse. While the total area of the springs is estimated to be less than 10 km
2, it is unknown how much of that is inhabited by the
Alcolapia. A planned contemporary soda ash mine would result in the construction of dikes and other infrastructure, as well as an increase in the local population. Such mines require large amounts of water, and any impact on water levels could impact the
Alcolapia populations that rely on the shallow springs [
36]. A hydropower dam proposed for the Ewaso Ngiro River in Kenya would further decrease the main source of water input to the lake from the northern side [
36], putting additional pressure on the southern freshwater springs.
The highest upper critical temperature (45.6 °C) recorded for a freshwater fish was for
Alcolapia grahami (Boulenger 1912), from nearby Lake Magadi in Kenya [
87], which is less than 3 °C warmer than the current upper range of the water temperature for the Natron Springs
Alcolapia. The only other cichlids known to naturally inhabit similar temperatures are the endemic cichlids from the genus
Danakilia, found in the shallow (<10 m) waters of Lake Abaeded in the Danakil Depression of Northern Eritrea and Lake Afrera in Ethiopia [
88], and
Iranochicla from Southern Iran [
89]. The water influx to the lakes inhabited by
Danakilia is also mainly from springs and ranges in temperature from 29–45 °C [
88]. Nevertheless, the
Danakilia, while abundant in lakes, are generally found in the coolest areas, and rarely in the upper extremes of water temperature, suggesting that while these cichlids may be thermal tolerant, if conditions allow the fish will reside in the cooler temperature range of their habitat. The projected air temperature increase of 2.5 °C (by 2050) from the GISS model for the Natron Springs
Alcolapia is consistent with contemporary surface temperature increases observed from ground monitoring stations (HadCRUT4 data). While the high specific heat capacity of water will protect the fish from abrupt temperature changes, we believe that projected climate change will subject the
Alcolapia to prolonged exposures to warmer conditions. As ectotherms, fish have poor insulation and their body temperature is predominantly controlled by the water body in which they live [
90]. Short-term positive impacts on growth of increases in environmental temperature are negated if the temperature surpasses the fish’s optimal range, beyond which growth and other necessary biological processes are compromised [
91]. Tropical fish, in general, experience narrower temperature fluctuations than temperate species and many already live close to their maximal thermal tolerance [
92,
93]. With climate change, fish are expected to be exposed to thermal regimes outside their thermal tolerance windows [
94].
Based on the 2006 the IUCN Red List assessment
A. alcalicus is considered endangered with a decreasing population trend. One of the primary causes listed for the imperilment of
A. alcalicus is climate change. Given that
A. ndalalani and
A. latilabris are more specialized than
A. alcalicus (
Table 1), we expect them to be more threatened by changes in their habitat [
95,
96]. Both
A. ndalalani and
A. latilabris are categorized as vulnerable by the IUCN Red List; the primary threat listed for both is habitat degradation. A projected decrease in precipitation and a potential decrease in the subsurface aquifer could lead to an overall loss of habitat extent or depth. All three species graze on algae and biocover growing on the rocks and stems of marginal plants. The marginal grasses also act as a nursery for juvenile fishes. A decrease in water depth in the springs could have devastating effects on the available habitat including a reduced amount of marginal vegetation (
Juncus sp.) that acts both as a vital habitat for the young cichlids, once released from the mouth of the female, cover from cannibalistic larger specimens, and extended surface area for algae and biocover. The projected population growth (average 102 inhabitants/km
2) by 2050 is further expected to stress the water resources of the arid region.
Our results also highlight the importance of the management of the local water resources at Chemka Springs. Chemka Springs contribute a major influx of water to the NYM dam reservoir. All natural vegetation except the Riparian zone immediately next to the springs have been removed (
Figure 2B and
Figure 7F). Palm plantations were observed to be encroaching on the springs. With an increase in population density, cropland, and agriculture (
Figure 10), the need for water abstraction also increases. Irrigated agriculture doubled in extent in the immediate vicinity of the springs; this is expected to further increase as rainfall and subsurface water may become inadequate for the growing needs of the population. Pollution from agricultural runoff may further adversely affect the limited habitat of undescribed
Ctenochromis sp. Due to potential fluvial habitat fragmentation and breaks in connectivity, it is unclear how much area would be available for the fishes to migrate if their current habitat at the mouth of the springs becomes uninhabitable. The site (springhead) is a popular swimming spot for both locals and tourists, but no formal management of the site exists to protect it from degradation, refuse left behind by visitors and other activities that are potentially detrimental to the habitat (e.g., washing vehicles in the shallow portions of the stream).
Arguably, the habitat of
O. hunteri in Lake Chala may undergo the least amount of change unless there is large water abstraction from Lake Chala, increased agricultural runoff, or introduction of additional species. The large volume of Lake Chala provides some protection against climate change. However, increased water abstraction due to climate change, introduced species and water quality degradation from sedimentation and agricultural runoff are important detrimental factors. Based on a 2006 assessment, the IUCN Red List categorizes
O. hunteri as critically endangered with a decreasing population. The primary threats listed by the IUCN are climate change and habitat degradation. Another threat to
O. hunteri is the introduction of species, especially
C. rendalli that compete directly for food and territory. The high reproduction rate of
C. rendalli also outnumbers the slower reproduction rate of the mouth brooding
Oreochromis. The wide range of tolerated environmental conditions, trophic adaptability and high reproductive rates have been shown to be predominant factors that have allowed tilapias (such as
C. rendalli) to become one of the groups of exotics fishes most widely distributed worldwide, exacerbating damage to native fish, and making them a primary cause of species endangerment and extinction [
97]. The voracity of these tilapias, especially during times of year where food resources are scarce leads to their ability to outcompete native fishes [
98].
One of the limitations in this study is the lack of historical or contemporary population data for the species examined. However, as described earlier, such data on African fishes outside the Great Lakes in small aquatic systems are non-existent for most of the endemic species. Therefore, our conclusions on the putative effects of environmental and local human induced stressors are based upon studies of other fishes worldwide undergoing similar stressors [
99,
100,
101,
102,
103,
104,
105] among many others. A well-studied example of a fish critically endangered by climate change and water abstraction for human uses is the Devil’s Hole pupfish (
Cyprinodon diabolis Wales, 1930). It lives in an underground cave with water from a single aquifer-fed thermal pool in the Desert National Wildlife Refuge Complex in Nye County, Nevada [
106]. By 2013, the wild population was recorded to have decreased to less than 40 individuals [
107]. It was among the first species to be listed on the U.S. Endangered Species Act in 1967 and was the focus of a legal battle over water rights, the culmination of which was a U.S. Supreme Court decision to cease groundwater abstraction for development. Despite this protection, populations of
C. diabolis have declined since 1995; one of the potential causes being increased water temperatures caused by climate change, and therefore, decreased dissolved oxygen. Projected climate change by 2050 was found to compromise the recruitment window for juvenile fish, decrease egg viability and increase larval mortality [
107].
While not examined here, increasing atmospheric deposition of sulphur, nitrogen, mercury and other compounds has been shown to have adverse effects on a variety of ecosystems including lakes and streams [
108,
109,
110,
111]. Future analyses should incorporate modeled rates of atmospheric pollutant deposition on the habitats.
Another recommendation is to increase the scope of the analyses to the continental scale with a focus on small aquatic habitats outside the Great Lakes. In a systematic review of studies examining predicted and observed effects of climate change on freshwater fishes, Reference [
112] found that tropical arid and semi-arid regions have been grossly understudied. Future predictions were heavily biased to the Nearctic (~75%), while the Palaearctic dominated studies examining the effect of recent climate change (~46%). Over half the published studies included at least one salmonid species, indicating a taxonomic bias. The recently recognized Ovalentaria [
113,
114] includes a number of the fish groups previously classified within the “perciformes” of previous authors. As currently recognized, the Ovalentaria includes an array of marine, and a few freshwater groups, among which the Cichlidae represent a major component of tropical freshwater diversity. Despite this taxonomic importance, Reference [
112] highlighted a general lack of studies from tropical freshwaters. Furthermore, less than 10% of the studies focused on fishes categorized as ‘
threatened with extinction’ by the IUCN. Based on the IUCN Red List version 3.1, of the species assessed, 1045 freshwater fish are considered critically endangered or endangered. Of those, climate change is a primary threat for 185 species; six are cichlids, with Cyprinidae accounting for 48.6%. When considering human-induced stressors such as agriculture, energy production, logging/wood harvesting, fishing, human intrusions, natural system modification, and pollution, 838 species have been listed as affected. Of these 143 (17.1%) are cichlids. When invasive species are considered as a major threat, 336 species have been assessed as critically endangered or endangered, of which 76 (22.6%) are cichlids. Native African species account for 33.1% of critically endangered or endangered species for which climate change, human-induced stressors, or invasive species are primary threats.