Application of Bayesian network inference algorithms for foodweb analysis: evaluating the impact of jellyfish predation on Irish Sea plankton

水母(Cnidaria的中上层海洋水母)是贪婪的浮游生物捕食者，可以在构建中上层(水柱)海洋生态系统中发挥主要作用，但由于使用传统的网捕技术对其进行采样存在困难，对其生态方面的描述仍然很少。我们建议使用最近发展的声学技术(多频率科学回声测深)来估计爱尔兰海以前被航空测量确定为根口“热点”区域的一群水母(桶形水母、章鱼根口水母、狮毛水母Cyanea spp.、海燕金丝水母和月亮水母Aurelia aurita)的丰度。我们还将使用垂直捕捞的渔网对非胶状浮游动物(如桡足类、两足类和水母捕食的鱼类幼体)进行采样，以便我们可以定量地描述不同水母丰度条件下浮游生物(水母‘食物’)的组成。我们预计，总的来说，在水母数量最多的地区，浮游动物的猎物将会减少，因为水母将捕获一些浮游动物。由于浮游动物在食物网中有许多可能的联系(浮游动物除了被水母消耗外，还可能相互消耗)，我们将应用强大的贝叶斯网络推理算法对多组区域自然变化的水母和浮游动物丰度数据进行推断，以推断最可能的食物网，从而推断水母对浮游生物群落的影响。这将极大地提高我们对水母捕食性影响的理解，并将提供对水母丰度增加可能产生的生态系统后果(例如对渔业招募的影响)的洞察，例如渔业驱动的鳍鱼减少或环境变化。有人提出，面对巨大的捕捞压力，水母会大量繁殖，因为捕鱼会把与水母竞争浮游生物的鱼赶走：鱼越少，水母的食物就越多。一旦我们有了一个强大的食物网模型，我们将用它与历史浮游动物丰度数据(由连续浮游动物记录器收集，这是一个长期的浮游动物采样计划，从商业船只上机会主义地运行)反向使用，根据其丰度的任何变化对浮游动物群落的影响，重建爱尔兰海水母丰度随时间的可能变化。爱尔兰海的鲱鱼种群在1972至1980年间大幅减少，在一种生态考古中，我们将在这段时间内浮游生物群落组成的变化中寻找线索，以增加水母的数量。尽管有人提出，水母可能会随着鳍鱼的减少而繁殖(这是所谓的“顺着食物网捕鱼”的结果)，但几乎没有直接证据表明这一点，因为水母丰度的时间序列很稀少(水母没有坚硬的部分，所以，例如，一旦死亡，它们在沉积物中几乎没有留下痕迹)。如果我们能够揭示水母丰度和鱼类丰度之间的历史联系，这对于寻求在生态系统背景下管理渔业的管理者来说将是一个非常有用的进步。该项目还将向广大生态学家观众展示推理算法在食物网分析中的力量。

Jellyfish (pelagic meduasea of Cnidaria) are voracious plankton predators that can play major roles in structuring pelagic (water-column) marine ecosystems, but aspects of their ecology remain poorly described because of difficulties associated with sampling them using traditional netting techniques. We propose using recently-developed acoustic techniques (multi-frequency scientific echsounding) to estimate abundance of a guild of jellyfish (the Barrel jellyfish, Rhizostoma octopus; the Lion's mane jellyfish Cyanea spp.; the Sea nettle Chrysaora hysoscella, and the Moon jellyfish Aurelia aurita) in a region of the Irish Sea identified previously by aerial surveys as a Rhizostoma 'hot spot'. We will also sample non-gelatinous zooplankton (eg copepods, amphipods and fish larvae that are prey for jellyfish) using vertically-fished nets so that we can describe quantitatively the composition of the plankton (jellyfish 'food') under conditions of varying jellyfish abundance. We expect, in general terms, that there will be fewer zooplankton prey items in regions where jellyfish abundance is highest because jellyfish will have captured some zooplankton. Because there are numerous possible links in the foodweb via which zooplankton could be consumed (zooplankton may consume each other in addition to being consumed by jellyfish), we will apply powerful Bayesian network inference algorithms to multiple sets of regionally-naturally-varying jellyfish abundance and zooplankton abundance data to infer the most likely foodweb, and thus the impact by jellyfish on the plankton community. This will greatly improve our understanding of the predatory impact of jellyfish, and will provide insight to possible ecosystem consequences (eg to fisheries recruitment) of increasing jellyfish abundance following, for example, fishery-driven finfish decline or environmental change. It has been suggested that jellyfish will proliferate in the face of high fishing pressure because fishing removes fish that are competitors with jellyfish for plankton: less fish means more food for jellyfish. Once we have a robust foodweb model, we will use it in reverse with historic zooplankton abundance data (collected by the Continuous Plankton Recorder, a long-term zooplankton sampling programme run opportunistically from commercial ships) to reconstruct the likely changes over time in jellyfish abundance in the Irish Sea from the impact any changes in their abundance would have had on the zooplankton community. Herring stocks in the Irish Sea collapsed between 1972 and 1980 and, in a kind of ecological archaeology, we will look for clues in possible changing plankton community composition over that time for increases in jellyfish abundance. Although it has been suggested that jellyfish may proliferate following finfish decline (a consequence of so-called 'fishing down the foodweb') there is little direct evidence of this because time-series of jellyfish abundance are scarce (jellyfish do not have hard parts so, for example, leave little trace in sediments once they die). If we are able to reveal an historic link between jellyfish abundance and fish abundance this will be a very useful advance for managers seeking to regulate fisheries in an ecosystem context. The project will also demonstrate to a wide ecologist audience the power of inference algorithms for foodweb analysis.