Discussion
We show that the success and failure of predatory biocontrol of schistosomes depends on key ecological mechanisms, namely resource competition, increased infectiousness with resource uptake, and size-dependent host mortality. These combined mechanisms result in overcompensation of parasite output to predator biocontrol on intermediate hosts and thus a unimodal response of ecological risk of human schistosome exposure. Compared to no predators (Fig. 1A & 2A), weak to moderate predator biocontrol performs increasingly worse (Fig. 1B & 2B) before eventually succeeding at extremely high levels of investment (Fig. 1D & Fig. 2D). Here, low predation pressure reduces resource competition and enables some juvenile snails to more rapidly grow, reach an invulnerable size, and produce more parasites, i.e. escape an otherwise strong juvenile bottleneck (de Roos and Persson 2013). Therefore, low predator stocking densities backfire because they only partially reduce host density, which relaxes competition for food and triggers a trophic cascade to facilitate higher parasite yields from fewer remaining hosts. In contrast, high predator densities kill more hosts to control parasite output, first by reducing infected snail lifespan below the prepatent, i.e. developmental period of schistosomes (~4 weeks), and then by extirpating snail populations (Fig. 1C & 2C). This result is consistent when predation is size-independent or categorically depends on host size (discrete host mortality ). The consistent pattern of backfiring until ultimate success exemplifies the challenges of ‘do or do not’ biocontrol by highlighting overcompensation based on resource competition as a primary driver.
Size-dependent predation drives parasite output by introducing a host size refuge from predation. We show when predators are gape limited (discrete host mortality ), biocontrol success depends on the upper limit of the binary host vulnerability window: there is a strong relationship between the upper limits to this window and a successful predator stocking density because predator gape limits shift the infected host size distribution to larger, more prolific hosts (Fig. 3, lighter curves). When predators kill almost all hosts (0–25 mm vulnerable; HCmax = 25 mm), they trigger host overcompensation and thus boost parasite output for both algae and detritus until eventually succeeding at high predator stocking densities (Fig. 3). In contrast, predator densities below this critical threshold not only underperform, but become increasingly worse and riskier the closer they approach it (Fig. 3, darker curves). However, when predators are unable to consume intermediate or large snail hosts, (HCmax = 5, 10, or 15 mm), total parasite production increases linearly. In other words, even at unrealistically high densities, predators that are small-snail specialists reduce resource competition, but do not eliminate large, prolific hosts. Therefore, if predator biocontrol programs fail to target large, disproportionately infectious hosts, they may be prone to backfire, regardless of management intensity.
In addition to a binary host vulnerability window, we also simulatedcontinuous host mortality , where predation risk follows a continuous negative-exponential function derived from experiments (Sokolow, Lafferty, and Kuris 2014). Under this scenario, amassing predator stocks fails to suppress parasite density until reaching unrealistic predator densities (Fig. 4), even when compared with densities simulated in a recent study that tested these prawns experimentally (dotted line; (Sokolow, Lafferty, and Kuris 2014)) and the maximum recommended density for aquaculture practice (dashed line; (Upstrom 1986)). Here, large snails are never completely invulnerable to predation, but experience a substantially lower risk than smaller individuals. Cumulative parasite output in these scenarios also stems from overcompensation. As seen previously, parasite output increases with resource supply rate (Civitello et al. 2018; Malishev and Civitello 2019) and overcompensation emerges under two conditions: whenever resource supply is high enough to support any parasite production and when the steepness of this response increases with supply rate. Combined, these results confirm how the aforementioned mechanisms shape overcompensation and further suggest resource type and supply exacerbate its role in size-dependent mortality under predator biocontrol.
Our model results contrast other simulations of predatory biocontrol of schistosomes (Sokolow et al. 2015; Hoover et al. 2019) that predict consistent success with increasing predator stocking density. Despite several differences between our individual-based model and these ordinary differential equation (ODE) models, one central difference drives these divergent forecasts: the positive effect of resource uptake on schistosome production by individual infected snails. Indeed, an extremely simplified ODE model of snail-schistosome dynamics illustrates how resource-dependent cercariae output can drive overcompensation due to increased background mortality (Civitello et al. 2018), supported by the SIDEB model. The positive effects of food quantity and quality are pervasive among schistosomes and other trematodes, infecting snails (Keas and Esch 1997; Civitello et al. 2018), suggesting a critical need to incorporate this general phenomenon when analysing disease mitigation strategies involving snails.
Given these divergent conclusions and the stakes involved in control intervention exacerbating or eliminating infection risk in vulnerable human populations, it is critical to evaluate existing evidence. Currently, only one published field study evaluates prawn-based biocontrol of human schistosomes (Sokolow et al. 2015). However, the study introduced an unspecified density of prawns into a water access point at one village in the Senegal River Basin and suggested this lowered schistosome infection in the following year compared to a control village not receiving prawns. While encouraging, these data cannot rigorously attribute this effect to the prawn manipulation, given the sample size of one per treatment and no follow up estimation of realized prawn density or survival. Despite a dearth of laboratory studies evaluating the population dynamics of snails and schistosomes in response to a range of prawn densities, one laboratory study shows gape-limited predation by prawns can relax competition in populations of uninfected snails, yielding fewer, larger individuals that produce many eggs, consistent with predictions from the SIDEB model (Sokolow, Lafferty, and Kuris 2014). Although this experiment did not incorporate infection, recent experiments on resource competition and food supply suggest these conditions could promote high parasite yields from the remaining larger hosts (Civitello et al. 2018, 2020). Data on predator densities, feeding selectivity, dynamics of snail population densities and size-structure, cercariae shedding, and human infection are needed to critically evaluate these alternative hypotheses and thus effective predator biocontrol of human schistosomes.