{"id":1921,"date":"2023-11-24T18:17:16","date_gmt":"2023-11-24T23:17:16","guid":{"rendered":"https:\/\/pressbooks.bccampus.ca\/ewemodel\/?post_type=chapter&p=1921"},"modified":"2026-04-08T19:13:03","modified_gmt":"2026-04-08T23:13:03","slug":"mediation-and-time-forcing","status":"publish","type":"chapter","link":"https:\/\/pressbooks.bccampus.ca\/ewemodel\/chapter\/mediation-and-time-forcing\/","title":{"raw":"Mediation","rendered":"Mediation"},"content":{"raw":"\"\"\r\n\r\nFigure 1. Schematic representing how tuna may mediate the feeding interaction between albatrosses and small pelagics<\/strong>\r\n\r\nThe development of the mediation concept and its inclusion in EwE comes from a model failure \u2013 that's often how we best learn how to improve models. There's nothing like failures! \u00a0Jim Kitchell and co-conspirators such as Carl Walters were at a NCEAS working group in Santa Barbara, Ca, making an Ecosim model of the central North Pacific Ocean with focus on tuna. \u00a0They modelled how the ecosystem of the North Pacific had changed over time, starting back before tuna fisheries were introduced in the open ocean. \u00a0Their model predicted that as the tuna fisheries increased, the target tuna populations would decrease, and this would have a cascading effect on their main prey, various small pelagic fish. \u00a0This, in turn would result in more prey being available for large piscivorous birds such as albatrosses, and the model indeed predicted an increase in these birds.\r\n\r\nEveryone was happy with this prediction \u2013 until they presented their model and findings at a seminar at NCEAS. There, a bird biologist chimed in: \"But that is not what has happened there, albatrosses declined with the introduction of tuna fisheries\",<\/em> and she made it clear that it was not because of incidental catch of seabirds in longline fisheries. Back to the drawing board then. Something was indeed wrong with the model, and the discussion soon focused on a non-trophic effect where the abundance of one group may impact the feeding interactions by other predators and their prey. We call this kind of indirect trophic effect \"mediation\".\r\n\r\nIn this case, tuna tend to stay in deeper and colder oceanic waters and only occasionally make foraging sprints up to the warmer surface layers. The smaller pelagics can better cope with the warmer surface water and when there are tuna around they tend to stay up there to minimize tuna predation.\u00a0 When the tuna do move up to forage, they drive the small pelagics closer to the surface. This in turn makes the small pelagics more accessible to albatrosses, which only can forage in the upper few meters of the ocean \u2013 they are not divers. \u00a0When tuna become less abundant because of fisheries pressure, there is less need for the small pelagics to stay in the upper water layers, indeed the risk of predation by birds that are better divers may well make them go deeper. \u00a0It's a balancing act of predation risk to be a small pelagic in a risky world.\r\n\r\nWhat we have here is a mediation effect where the abundance of one group (tuna) impacts the feeding interaction between two other groups, albatrosses as predators and small pelagics as prey. \u00a0An example of how to implement this in EwE is illustrated in Figure 2.\r\n\r\n\"\"\r\n\r\nFigure 2. Mediation function for how tuna may impact the feeding interactions between albatrosses and small pelagics. X axis is tuna biomass, Y axis is relative exposure or vulnerability of small pelagics to albatross.<\/strong>\r\n\r\nIn Figure 2, the X-axis represents the biomass of the mediator, tuna. When setting up mediation (Ecosim > Input > Mediation)<\/em> one can define a shape such as in Figure 2. The next step then is to define the mediating group (Define mediating groups and fleets <\/em>on the same form), so that the X-axis is defined once the shape has been set. \u00a0The Y-axis represents the feeding interaction between the impacted predator and prey, which is defined at Ecosim > Input > Mediation > Apply mediation<\/em> and which can be applied to consumers and producers separately[footnote]In principle, this can also be applied to fleets, but we have not implemented that yet. Ask if you need it.[\/footnote]. \u00a0In the original case, one would select tuna as the mediator, and as impacted groups, albatrosses as the predator and small pelagics as the prey.\r\n\r\nThe blue vertical stippled line in Figure 2 represents the Ecopath base mediator (e.g., tuna) biomass, and it can be moved freely. Moving it far to the left would imply that lower tuna biomasses would not result in notable less interactions between the impacts groups, but higher would. So, that would not work for the original case, which is probably best represented where the line is now since the mediating species is not expected to ever grow much higher in biomass than its ecopath base value.\r\n\r\nWhat shape should one use for a given mediation function? \u00a0There is no clear answer to that, so you should evaluate several alternatives. This is not an issue that has seen much attention in field studies, since it is typically difficult to obtain data well spaced over a range of mediator biomass densities, though the interest is growing thanks to the implementation in EwE. \u00a0The best advice is to do what makes sense, and to try alternative formulations to evaluate their impact notably as part of time series fitting. Or, with a warning, \"handle mediation functions with care\".[footnote]Harvey, C.J., 2014. Mediation functions in Ecopath with Ecosim: Handle with care. Canadian Journal of Fisheries and Aquatic Sciences 71, 1020\u20131029. https:\/\/cdnsciencepub.com\/doi\/10.1139\/cjfas-2013-0594<\/a>[\/footnote]\r\n
\r\n\r\nWhat do you do if the (time-varying) mediator isn't a functional group or fleet in your model?\r\n\r\nMake it a group. You can for instance add a detritus group that isn't connected to anything, and then force the \"biomass\" of that group with the time-varying pattern.\r\n\r\n<\/div>\r\nMediation functions are used in both Ecosim and Ecospace. \u00a0The mediation functions are implemented in the code the same way as for forcing functions (see Environmental impacts<\/a> chapter). Thus, when you apply the forcing functions to a consumer, you can (as for environmental forcing functions) choose between applying to search rate (a<\/em>), vulnerability (v<\/em>), arena area (A<\/em>), or both v<\/em> and A<\/em>, combined. \u00a0When applying mediation to a producer, the production rate (P\/B<\/em>) is impacted.\r\n
\r\n\r\nMediation can for instance be used to model,\r\n