Long-term collapse in fruit availability threatens Central African forest megafauna

Fruit decline threatens forest elephants Large mammal herbivores in African tropical forests are major consumers of fruit, and many tree species rely on these consumers for dispersal of their seeds. Bush et al. monitored fruit production over three decades in a protected national park in Gabon, showing an 80% decline across the 73 plant species monitored. At the same time, photographic records of forest elephants over the past decade indicate a substantial decline in body condition in these major herbivores. These results suggest that the capacity of the ecosystem to support the elephant population is decreasing, a worrying prospect in an environment that is still protected from other threats such as hunting and deforestation. Science, this issue p. 1219 Plant reproduction has slowed and forest elephant body condition has declined over recent decades in a forest reserve in Gabon. Afrotropical forests host much of the world’s remaining megafauna, although these animals are confined to areas where direct human influences are low. We used a rare long-term dataset of tree reproduction and a photographic database of forest elephants to assess food availability and body condition of an emblematic megafauna species at Lopé National Park, Gabon. Our analysis reveals an 81% decline in fruiting over a 32-year period (1986–2018) and an 11% decline in body condition of fruit-dependent forest elephants from 2008 to 2018. Fruit famine in one of the last strongholds for African forest elephants should raise concern about the ability of this species and other fruit-dependent megafauna to persist in the long term, with potential consequences for broader ecosystem and biosphere functioning.

T he largest plants and animals on the planet are disproportionately important for the metabolism and functioning of our ecosystems (1). However, they are also more susceptible to extinction (2), and "global downsizing" through loss of megafauna and megaflora is likely to have negative consequences for the biosphere (1). The African tropics are home to much of the world's remaining wild megafauna, but even here they are mainly confined to areas where direct human influences are low (3,4). Securing the future of these megafauna populations will depend on our ability to protect against human threats, such as habitat loss and hunting for international trade (5,6), as well as to maintain the health and productivity of the habitats that support these species. The creeping influence of the Anthropocene means that rapid climate and atmospheric changes will be felt even where direct human pressures are low, with potential for far-reaching impacts on habitats and species (7). Detecting, quantifying, and understanding changes to the health and functioning of the remaining safe havens for threatened megafauna should be a key conservation priority during this time of rapid change. In this study, we used a rare long-term dataset of plant reproduction and a photographic database of forest elephants to assess food availability and body condition of an emblematic species of megafauna within a Central African rainforest.
Fruit is a keystone resource for many of the African megafauna (8)(9)(10). Fruit production is highly dependent on climatic cues, and there is abundant evidence that climate change has already caused shifts in the timing of plant reproduction in temperate regions (11)(12)(13). Data on reproduction of tropical plants are generally scarce or have not been recorded for long enough to track such changes (14). However, tropical plants will be affected by climate change, and changes in tropical tree reproduc-tion have been detected at several sites for which long-term data are available (15)(16)(17). The mechanistic causes of these changes are not universal and are mostly unknown. The critical minimum temperature hypothesis describes how, in some Afrotropical plant species, flowering is not triggered until temperatures drop below a certain threshold (18). For these species, we would expect reproduction to be cued less often as temperatures rise.
Lopé National Park, Gabon, is a nationally designated protected area and a UNESCO (United Nations Educational, Scientific and Cultural Organization) World Heritage site and is a relative safe haven for Central African megafauna, such as great apes and forest elephants (3,4,19). It also hosts the longest continuous study of tropical tree phenology in Africa (20). From 1986 to the present, researchers at the site have monitored tree species that are important to the diet of gorillas, chimpanzees, and elephants. Once a month, focal tree crowns are observed from the ground (via binoculars), and the proportions of each canopy that are covered with flowers, fruit, and leaves are recorded (21). We used these data to quantify changes in the probability of encountering flowers, unripe fruit, and ripe fruit for 73 species over a 32-year period (1986-2018; n = 260,431 monthly crown observations, n = 2007 focal tree crowns; tables S1 and S2), while accounting for individual and species-level variation by using generalized linear mixed models (GLMMs) with a binomial error structure (22). We replicated this analysis for a subset of species (n = 14 species) that bear fruit previously identified as especially important in the diet of forest elephants during a 30-month dung study and an 8-year observational study of elephant diet at the site (9). We also calculated a ripe fruit availability score for each year as a proportion of maximum theoretical fruit availability (equivalent to all trees of all species bearing 100% canopy cover of ripe fruit for 12 months of the year) by selecting species that had been monitored continuously throughout the study period (n = 40 species) and, separately, a subset of these species that are especially important components of the elephant diet (n = 7 species).
We found that trees at Lopé are reproducing less often and that the probability of encountering flowers and fruit has declined significantly over time (GLMM; tables S3 and S4 and fig. S1). The average rate of encountering ripe fruit for all species (n = 73) dropped from 1 in every 10 trees in any given month in 1987 to fewer than 1 in 50 by 2018, a substantial 80.9% decline ( Fig. 1A and table S4). Although there was some variation between species, all species declined except one (Dacryodes buettneri; table S5 and fig. S2). Encounters of ripe fruit important for elephants (n = 14) declined at an even faster rate (−87.8%), from one in five trees in 1987 to fewer than 1 in 40 in 2018 (Fig. 1A). A general reduction in fruit availability for species monitored throughout the time period demonstrates that larger fruiting events have not compensated for the diminished frequency of fruit encounters (Fig. 1B). Availability of fruit important for elephants dropped steeply between 2000 and 2003, and the yields of the best years for fruit production after 2004 were lower than those of the poorproduction years before 2000 (Fig. 1B). Although the long dry season (June to September) has always been a time of fruit scarcity (10), historical seasonality in fruit availability has disappeared in recent years, owing to a major reduction in ripe fruit observed from October to March, which was previously the season of abundance (Fig. 1C).
Our results showing declines in the probability of encountering flowers and unripe fruit (fig. S1) indicate that suppressed production of ripe fruit is not primarily a pollination or fruit maturation issue. Nor is it likely to be due to any negative consequences of tree senescence, because we removed trees that died or were diseased prior to analysis and found no evidence that tree size (as a proxy for tree age) influenced reproduction over time ( fig. S3). Instead we consider it likely that climate changes experienced at the site have contributed to this shift in reproduction. Global warming has caused minimum daily temperature to increase, on average, by 0.25°C per decade at Lopé (23), which may be a key factor in reduced reproduction for some tree species that rely on a critical minimum temperature to trigger flowering (18). Rainfall has simultaneously decreased at the site by 75 mm per decade (23); thus, all tree species might be suffering because of water stress (24).
Given the decline in fruit since 1986, it is likely that fruit-dependent wildlife-such as forest elephants, great apes, monkeys, and many bird species-has been affected. Longterm population data are not available for these species at Lopé. Other measures of population health, such as body condition, can be used to gauge population responses to environmental change over relatively short periods. Forest elephants, the largest frugivores in the ecosystem (9), have been consistently photographed by researchers and visitors to the site since the late 1990s, resulting in a large photographic database (>80,000 photos). We used this database to evaluate annual and seasonal trends in the external body condition of forest elephants between 1997 and 2018, hypothesizing that elephant body condition has declined along with reduced food availability. Elephant body condition in photos was scored systematically by means of a custom-built web application and user interface (22). Scorers (n = 6 individuals) did not have access to the time and date of photographs, nor to the research question and hypothesis. Scoring effort (number of photos viewed per scorer) varied, but we found high agreement among scorers who used a standardized test database (mean intraclass correlation coefficient of 0.89, n = 200 photos). A total of 2823 photos met the strict imagequality criteria for scoring, and we used linear mixed effects models (LMMs) to quantify changes in elephant body condition (accounting for elephant age) over the full 21-year period, as well as separately for the first 11 years (1997 to 2007) and last 11 years (2008 to 2018) of data.
We detected long-term declines in forest elephant body condition at Lopé (LMM). For the period between 1997 and 2018, mean body condition of the population declined by 5.0% for all age classes. However, uncertainty was high, ranging from a small improvement in body condition to a large decline [LMM; 95% confidence interval (CI) = +0.9 to −11.2%; table S6]. The change was most pronounced in the second half of the time period (2008-2018) when body condition declined, on average, by 11.1% (LMM; 95% CI = −4.3 to −15.6%; Fig. 2A and table S6). Body condition varied seasonally in the first half of the time period (1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007), appearing to track fruit availability (dipping in June and peaking after the long dry season) without any obvious lag at the monthly resolution of this analysis (Fig. 1B). The sharp drop in body condition in November during 1997-2007 was inconsistent with fruit availability, but closer inspection of the data indicated that this finding was highly influenced by one very thin individual (body condition score = 2), recorded in November 1999, out of only three photos available for this calendar month during this period (table S7). In the 2008-2018 period, the peak in body condition after the long dry season had disappeared ( Fig. 2B and table S6). However, the sparse data for 1997-2007 and high uncertainty in the CIs for seasonality in body  condition mean that comparisons between the early and late time periods should be made with caution (tables S6 and S7). It is not known whether the changes observed in body condition in this study have affected forest elephant population health or dynamics in the study area. However, studies of African savanna elephants show that environmental stressors can have substantial longterm consequences for both individual fitness and population dynamics, with reproductive females and calves particularly affected (25). Reduced food availability could also act in synergy with other factors (such as disease) to magnify negative physiological consequences (table S8). Although the biological mechanisms and consequences of declining body condition are unclear at this point, the effects on forest elephant populations across the region are unlikely to be benign, particularly when coupled with illegal hunting, habitat loss, and habitat degradation (3).
These declines in both plant reproduction and elephant body condition are indicative of system-wide change and are expected to have disproportionate impacts on the functioning and metabolism of the ecosystem. A reduction or displacement of historic populations of large frugivores in this region, along with diminished availability of seeds, could lead to collapse of seed dispersal (8), landscape-level shifts in habitat structure (26), reduction of carbon stocks (27), and potential for increased competition with humans for food (28). However, long-lived plant and animal species are able to buffer environmental change to a certain extent (1), and the broad diet of forest elephants (9) means that alternative food sources could offer some relief. Nonethe-less, environmental stress can have lifelong negative impacts on elephant survival and fitness (25), and the impacts of these changes should be closely monitored.
The long-term plant and animal datasets presented here are rare for the tropics, and it is possible that such changes may be occurring elsewhere, undocumented. These data serve as a reminder that even where direct human pressures are low, plant and animal communities may not be protected from the creeping influences of the Anthropocene. Coordinated international efforts to relieve direct human pressures and to halt and reverse climate change will be critical to saving the remaining megafauna and megaflora of the African tropics and preserving their specialized roles in the functioning of our biosphere.