Examining the Extinction of the Barbary Lion and Its Implications for Felid Conservation

Estimations of species extinction dates are rarely definitive, yet declarations of extinction or extirpation are important as they define when conservation efforts may cease. Erroneous declarations of extinctions not only destabilize conservation efforts but also corrode local community support. Mismatches in perceptions by the scientific and local communities risk undermining sensitive, but important partnerships. We examine observations relating to the decline and extinction of Barbary lions in North Africa. Whilst the extinction predates the era of the scientific conservation movement, the decline is relatively well documented in historical records. Recently unearthed accounts suggest Barbary lions survived later than previously assumed. We use probabilistic methods to estimate a more recent extinction date for the subspecies. The evidence presented for a much later persistence of lions in North Africa, including generations when sightings were nil, suggests caution when considering felid populations as extinct in the wild. The case raises the possibility that captive animals descended from the Moroccan royal collection are closer contemporaries to wild Barbary lions. Furthermore, our results highlight the vulnerability of very small lion populations and the significance of continued conservation of remnant lion populations in Central and West Africa.


Introduction
Several statements of extinction relating to felids have been proven unreliable, such as the recent rediscovery of the Barbary leopard [1] and the late persistence of the Caspian tiger, recognized as extinct since the early 1970s, yet later found present in Turkey where a local trade in hunted skins persisted into the 1980s suggesting survival into at least the early 1990s [2]. In both cases local people's observations remained unknown to science and therefore had no impact on conservation policy.
Wild populations of lion (Panthera leo), like other large mammalian carnivores, are suffering severe decline and in Africa have contracted sharply over the past 50 years [3], [4]. Outside East Africa, lion populations are fragmented, with remnants in Central and Western Africa threatened with extinction [5], [6]. The present-day status in sub-Saharan Africa mirrors the situation north of the Sahara a century ago where the once extensive lion distribution from North Africa to India was reduced to fragmented populations by the 20 th century [7].
The North African 'Barbary lion' or 'Atlas lion' occupied the Maghreb, the region isolated from the rest of non-arid Africa by the Sahara [8]. Until the 18 th century, Barbary lions ranged from the Atlas Mountains to the Mediterranean [7]. Extensive persecution in the 19 th century reduced populations to remnants in Morocco in the west, and Algeria and Tunisia further east, all of which were extirpated during the 20 th century [9], [10].
The Barbary lion (P. l. leo) is considered distinct from the six commonly defined sub-species in the rest of Africa [11], owing to its geographic separation, morphology and unique montane habitat with cold winters. Although the phylogenetic status of lion populations remains unclear [12], [13], recent morphological and genetic studies consistently differentiate northern lions (India to North Africa) from sub-Saharan African lions [9], [14], [15].
The IUCN's assessments of species status, threats and extinctions have been established over the past four decades [4], which post-dates the extinction of the Barbary lion, so Red List declarations on the sub-species are based on reviews of literature [11]. The IUCN Wild Cats Status Survey and Action Plan recognized that certain captive lions may be descendants of the Barbary lion, designating the subspecies as 'extinct in the wild' [11]. Currently all extant African populations are grouped together as a single taxon, Panthera leo, with the species listed as 'regionally extinct' across its former northern range [16]. Only the Indian population is recognized as a distinct sub-species, P. l. persica [17], whilst North African populations (i.e. the Barbary lion) are mentioned in the context of both the IUCN's Indian and African lion accounts [4].
In the 16-18 th centuries many accounts reported lions in the western Maghreb (northern Morocco) near the Atlantic and Mediterranean coasts [7], [18]. Up to the 1830s lions were still seen in these coastal areas, the Rif mountains and the Marmora forest [19], however records remain sparse throughout the 19 th century [20]. By 1880 lions had retreated south of the Bou Regreg and Taza passes [19] into the Atlas Mountains and areas bordering the Sahara, where human populations were largely nomadic. Previous commentators suggested Barbary lions were extirpated sometime between 1920 and 1930 [7], [19], [20]. Later sightings, however, have since been documented with the last in the High Atlas in 1942 [18].
In the eastern Maghreb (Algeria and Tunisia) lions frequented coastal-forested regions, the Tell Atlas and the Aurès mountain ranges. By the late 1800s, Tunisian sightings were confined to localities adjacent to the Algerian provinces of Souk Arras and Tebessa. Although no lions were shot in Tunisia after 1891 [7], rumors of their survival persisted in the 1900s in the Khmir Mountains and near Feriana [21]. The population was probably contiguous across the border, corresponding to home range sizes equivalent to those observed in present-day sub-Saharan Africa and India [22], [23], [24]. In Algeria, lions persisted into the 1890s and hunting accounts, the capture of wild cubs and photographs of tame lions were still widely reported [7], [20]. These observations, alongside interview evidence, suggest that a small population survived in Algeria well into the 20 th century [25], long after 1893 when Algeria's supposedly last lion was shot [7].
Several reviews have considered the history of sightings [7], [19], [20], [26], [27] and the IUCN recognizes that lions persisted in Morocco into the 1940s [17]. Nevertheless these reviews have inadvertently missed, ignored or have lacked access to many local sources and literature accounts. A systematic evaluation of the last sightings in the Maghreb will provide more complete insights into persistence of remnant lion populations and the resilience of large mammalian carnivores to human pressure.
Several menageries in Europe held Barbary lions in medieval times [9] and they were popular exhibits in public zoological gardens in the 1800s [7], [20]. By the early 1900s zoos and circuses in Europe and North America often promoted their lions as ''Barbary'' [13], although true representatives were said to be only found in the collection of the Sultan of Morocco, derived from animals caught by local tribes [20]. The significance of this collection was not recognized until the 1970s after the lions were moved from the Royal Palace, Rabat, to a new zoo at Temara when a study identified animals with physical characteristics of the Barbary lion [20]. Despite several attempts, a formal scientific breeding program is yet to be established (Frey pers. comm.). Nevertheless, captive breeding has experienced a recent renaissance [28], [29] and a studbook for these animals (hereafter 'Moroccan Royal lions') has been developed [13].
Debate surrounds the authenticity of Moroccan Royal lions as descendants of wild Barbary lions [13]. One concern is that Moroccan Royal lions hybridized with sub-Saharan African lions potentially introduced to the collection before the 1970s [20]. Definitive genetic matches have not yet been established between Moroccan Royal lions and wild Barbary lions [13]. The few studies of Barbary lions that utilize museum samples [14] are limited by the scarcity of wild-origin reference specimens. Most recent genetic studies [30], [31], [32] rely on Royal lion samples from two zoos, covering at most 4 (possibly only 2) of the 12 maternal bloodlines [13]. In the absence of genetic data, owing to the lack of verified wild specimens, and working on the precautionary presumption that Barbary lions are taxonomically distinct from other lions, assessment of purity relies on 'pedigree information' which at best can be taken from the recently established studbook of lions of known ancestry arising from the Moroccan Royal Collection [13].
The North African-Asian population of lions [14] is only represented by today's Asiatic lion (c.350 wild individuals and c.100 zoo captives), so the potential significance of captive Moroccan Royal lions (c.90 individuals) is not trivial [13]. In the absence of definitive genetic comparisons we examine the generational separation between lions surveyed in Temara Zoo in 1974 and wild-caught Barbary ancestors. This infers the degree of opportunity for hybridization had non-Barbary individuals bred within the Royal collection, and the potential purity and relevance of the extant captive population.
Our study re-examines the historical decline and extirpation of lions in North Africa. We calculate likely extinction dates for different populations and thus patterns of decline. We also examine new estimates of generation time in the captive population originally derived from wild North African ancestors and thus the likelihood of introgression with sub-Saharan individuals.

Sightings Information
Existing literature was reviewed for accounts of sightings and kills by date and location, including documented interviews in addition to recent interviews by the authors (Tables 1-6). The collation includes sightings, photographs, accounts and recollections since 1839. Observations have been taken from recorded accounts from first-hand sources (written recordings of interviews), from direct interviews of eyewitnesses or a first-hand recollection of an eyewitness account; all sightings in the dataset are therefore from second-hand sources or better.
Morocco and the Morocco-Algeria border are considered as the western Maghreb, whilst the eastern Maghreb covers Tunisia and northern Algeria. The meridian bisects sightings west and east, the most proximal being separated by 220 km of desert and temporally by nearly 70 years. The nearest contemporaneous sightings across this divide occurred in 1912 and are separated by a distance of over 450 km. A small central population in the Saharan Atlas (observed in 1898, 1912, 1920 and 1935) possibly traversed both regions but, owing to the size of the dataset, exclusion of these 3 sightings made very little difference.

Inferring Extinction Dates
While the last sighting of a species is often used as the time of extinction, it rarely corresponds to the true extinction date [33].
Following Solow [33], let sighting times in years be order t 1 ,t 2 ,…,t n , where t 1 = 0. Sampling from the uniform distribution, the unbiased estimate of extinction time is and the expected year of extinction is T E plus the year of t 1 . The where a = 0.05 (after [33]). Several other methods of inferring dates of extinction have been developed (see Solow [33] for a review). However, in a study of putatively extinct North American birds, Vogel et al. [34] showed that the simple Poisson process model, used here, had as good or better fit than other more complex models, such as the truncated exponential and Weibull.
Here we calculate the extinction dates for sightings of lions in the Western (Table 1; n = 16) and Eastern Maghreb (Tables 2-6; n = 66).

Testing Behavioral Change
We tested whether lions appear to change their behavior as populations decreased in size to operate singly ('09) rather than in groups ('19) as a binary set of sightings [35].
In a series of N sightings, X 1 , X 2 ,…, X N , each sighting X i is coded as X i = 1 or X i = 0. Of the N sightings, let m equal the number of sightings of lions in groups and n equal the number of sightings of lions singly. Then and n = N-m. The cumulative number of group sightings at each point is then determined. This frequency is where j = 1,2,…,N.
The statistic for testing the hypothesis of change is The expression is evaluated for all values of j from 1 to N-1. D m,n is the largest absolute difference observed in the sequence. We may reject the H o at 0.05 if D m,n §1: 36 ffiffiffiffiffiffiffiffiffiffiffiffiffi N=mn p Here we test the null hypothesis that there has been no change in the probability of sightings of lions in groups over time in the Western and Eastern populations.

Estimation of Generations between Moroccan Royal Lions and Wild-caught Specimens
Hemmer [20] suggested that 20 generations separated Moroccan Royal lions surveyed in 1974 from wild lions captured c.1899, suggesting a 3.75-year generation time and multiple opportunities for hybridization with non-Barbary lions. The average age at which a wild female lion gives birth is 6.5-8 years [11], [36]. In the Moroccan Royal lion studbook, the mean generation time through to today's juveniles is 7.4 years (STD63.2) in a range of 6.6-10 years [37]. The 30 breeding females since 1974 have a mean age at first surviving litter of 6.3 years (STD62.9). These observations suggest that Moroccan Royal lions have a generation time similar to the birthing age for wild lions, 6.5 years [36]. We calculate the number of generations between the animals moved to Temara Zoo and the likely last wild-caught animals, based on a calculated inferred extinction date. We also calculate the number of generations occurring between sightings of wild Barbary lions as an indicator of persistence.

Sightings of Lions in North Africa
The list of recent historical lion sightings in North Africa includes data for the western Maghreb from 1840 to the 1940s (Table 1) and the eastern Maghreb from the 1830s to the 1950s (Tables 2-6). Sightings become infrequent in the eastern Maghreb after the 1890s and in the western Maghreb after the 1920s. Our examination of historical accounts reveals that lions occupied the Saharan Atlas (Fig. 1), much further south than previously reported by Schnitzler [27]. The precise location of the aerial photograph on the Casablanca-Dakar route (Fig. 2) is unknown, however our research has identified that it was taken in 1925 when flights commenced [38]. A postcard edition of the image has recently been discovered with the caption ''Un lion photographié en avion dans l'Atlas'' (Fig. 3). Since this predates discussions on the extermination of lions in the region [19], its significance at that time was then unknown. The photograph by Flandrin is the last known image of a wild Barbary lion.

Recent Extinction of Lions in North Africa
In recent years it has become widely asserted that the animal shot in 1942 on the Tizi-n-Tichka pass in Morocco's High Atlas Mountains [18] has been considered the last wild Barbary lion [4]. However, our analysis suggests that wild lions actually persisted longer in Algeria until 1958 (T E ), with an upper bound for the 95% CI of T u E = 1962. This is 10 years after 1948 (T E ), the estimated extinction date of the western (Morocco) population. However, the estimated 95% confidence intervals have a substantial overlap: with upper bounds (T E ) of 1962 for the

Change in Behavior
The frequency of observations, particularly after the 1920s, which involves solitary animals rather than groups (Tables 1-6), may be entirely due to a dwindling population with lower density or a potential change in social behavior (from group to solitary living). Owing to the uncertainty surrounding some sightings, and therefore the order of occurrence of single and group sightings of lions, we analyzed the date at the two extremes; all groups occurring as late as possible and then as early as possible (only relevant to the eastern population). For the western population D m,n = 0.351 compared with a critical value of 0.658 (N = 18), so that we cannot reject the H o and suggest that there has been no change in behavior. Likewise for the eastern population no change in behavior was detected even though the inhabited areas are adjacent to human settlements; D m,n ranged from 0.235 to 0.249 compared to a critical value for both of 0.250 (N = 121).

Persistence of Lions
If we consider the gap size between sightings, lions appear able to persist whilst unseen by humans for at least a generation, if not two. A more rigorous analysis of gaps sizes for sightings is not possible due to ambiguity in some sighting dates. In the western Maghreb lions went unseen for 10 years, reducing to 3-5 years towards the end of the record. In the eastern Maghreb the gap sizes between later sightings were generally greater than 5 years, although precision is difficult owing to uncertainty in some dates. In one case the gap could be .10 years (1900s to 1910-1912).

Generations between Moroccan Royal Lions and Wild-caught Specimens
Current breeding animals in the Moroccan Royal Lion studbook [13] are separated from the Royal Palace animals by just four generations (in one case, three generations) and recent cubs [39] are five generations removed. Hemmer's [20] estimate of generations between wild ancestors and the lions in the Royal collection in the 1970s, would suggest that today's surviving descendants stand 25 generations from wild ancestors (Fig. 4). However, assuming a generation time of 6.5 years [36], our analysis places today's Moroccan Royal lions as much closer descendants; with the youngest breeders only 16 generations removed from wild ancestors (Fig. 4).

Discussion
Our analysis suggests that relatively recent sightings of lions (1940s and 1950s) are not exceptional. Whilst we corroborate recognized sightings in Morocco in the 1940s, we also suggest that lions persisted in Algeria into the late 1950s, nearly seventy years later than considered in previous reviews. Barbary lions experienced a combination of factors of decline consistent with the threats to modern-day felids; habitat availability, wild prey availability, livestock husbandry and management, human behaviour, land use and socio-economics [40], [41].
A more recent extinction date for Barbary lions raises several questions: (a) how lions managed to persist in degraded North African ecosystems; (b) whether later persistence in North Africa provides insights for conserving marginal lion populations in West and Central Africa; (c) whether wild Barbary lions were more recently taken into captivity, making current animals closer descendants than previously considered.

Recent Persistence of Lions in North Africa
Previous assumptions that lions have been absent from North African ecosystems for most of the past century is challenged by our analysis. We suggest that small populations persisted largely undetected by humans for several generations, indeed many decades. In Morocco, an inhospitable interior always made encounters with lions rare, whilst in Algeria sightings across the population's range occurred on an almost annual basis up to 1894, although thereafter less frequently than every 1 lion generation (6-7 years). Our calculation of the time of extinction (T E ) corresponds with the recent suggestion that the last Barbary lion was probably lost in the destruction of forests north of Setif in 1958 during the French-Algerian War [25].

Adaptation of Lions in North African Ecosystems
Up to the late 1800s, hunters reported lions traversing from northwest Algeria, westwards into Morocco and from northeastern Algeria eastwards into Tunisia [19]. After the 1880s, the pattern of sightings suggest that lion populations retreated broadly in two directions; in Morocco southwards away from coastal regions through the Rif, Middle and High Atlas Mountains and the Saharan fringes; and in Algeria eastwards into the Tell Atlas and the Aurès Mountains bordering Tunisia. In both regions small populations survived at low densities in remote areas for several generations. Literature and oral accounts suggest that lions persisted through certain behavioral adaptations (hunting domestic livestock, engaging in nocturnal activity, living in small groups or pairs) and shifts in range (leaving deforested localities, moving to outlying areas and higher altitudes, and following water points in arid regions). Many of these particular behavioral adaptations have since been observed in contemporary populations of lions in human-dominated landscapes in sub-Saharan Africa [42], [43] as well as in restricted available habitat in India [11].
Our analysis of historical records in the Maghreb suggests that lions appear not to adapt their social behavior as a response to human habitat pressure. There was no significant change from group to solitary living by lions in the remote western Maghreb regions, whilst importantly even the eastern Maghreb lion populations did not exhibit a significant change in group living in the final decades of their existence, despite increasing encroachment by human populations. Although lion population density is typically lower in desert and semi-desert areas, pride size generally remains comparable in desert ecosystems and moister ecosystems with higher lion density [22]. Interestingly, human presence has been shown to provide advantages to lion cub survival by   (AD 1900(AD -1960. Grey shading indicates Mediterranean scrubland ecosystems [8]. Circular markers indicate sightings in western Maghreb (8-21 in Table 1); triangular markers indicate sightings in eastern Maghreb (134-149) from incidents described in Table 6   removing secondary predators [44], and this may have enabled lion persistence in the Maghreb. Final extirpation of lions through hunting was a response to livestock predation; a prey dependency probably driven by reduced habitat and fewer wild ungulates [18]. This final collapse may have been exacerbated by the species' behavioral need to remain in social groups with proportionally greater local resource requirements. Group living has been shown to be similarly maintained by lion populations in semi-arid environments such as the Kalahari and Etosha [45] [46]. In contrast, leopards (Panthera pardus) still persist at lower population densities in Morocco [1].

Heritage of Moroccan Royal Lions
Moroccan Royal Lions have been described as ''an obvious relic of the original Barbary lion gene pool'' [20]. Our analysis suggests fewer post-wild generations have occurred during the period where this population has existed solely in captivity. Today's lions, descended from the Moroccan Royal collection, could be closer relatives to wild ancestors than previously considered. In the absence of definitive nuclear DNA profiles from wild Barbary lion specimens, the precautionary principle suggests that Moroccan Royal lions should be conserved as descendants of the Barbary lion until science can tell us otherwise. The Moroccan Royal lions offer one of the few scenarios in which restoration of lions into regions where the species is long extinct could be envisioned as having useful conservation value [47].

Conclusions
Insights from historical sightings are relevant to current lion conservation. We suggest that wild lions persisted in the Maghreb into the 1950s, much later than previously recognized. The lion is a well-known, visible and potentially threatening species, yet small populations survived in North Africa decades after being generally considered extinct. This persistence reflects the recent rediscovery of a small population of Barbary leopard nearly 20 years after the last previous sighting and a decade since being declared extinct [1] [48]. Careful consideration should be given to mammalian carnivores currently presumed extinct or near-extinct in other regions, coupled with a greater understanding of extinction patterns and the conservation potential in relict populations.
Finally, we suggest caution when considering the current conservation status of lions. Although lions in the Maghreb adapted to reduced population density, prey availability and habitat encroachment, our analysis reveals that lion group-living behavior did not change significantly as human pressures increased. As a pride-forming species [45], P. leo populations are prone to collapse, whereas other felids may survive at lower local population densities by not living in social groups [49]. Lions in today's small populations in Central and West Africa persist [16] [50], even if rarely seen, in fragmented remnants, yet clearly exist at the edge of a precipitous drop into extinction. Continued, carefully considered conservation effort remains vitally important.