The main objective of the research in this thesis was to understand the relative importanceof top-down and bottom-up forces on the functioning of tropical intertidal seagrassmeadows. Seagrass beds play an important ecosystem-structuring role through theircapacity to alter abiotic conditions creating strong bottom-up effects. At the same time,predators can also exert crucial community structuring top-down effects through trophiccascades in these ecosystems, which suggests that the interplay between top-down andbottom-up effects may be essential for the functioning of these ecosystems. Key questionsin this thesis were: How can an intertidal seagrass ecosystem function under the constantrisk of biochemical stress conditions (high sulfide levels, bottom-up effects). What is thefunction of mutualistic interactions (bottom-up effects)? What is the role of a migranttop-predator (top-down effects)? And finally, do bottom-up and top-down effects havesynergistic impacts on ecosystem functioning and does this give us better insight into howseagrass ecosystems may respond to possible enhanced environmental stress condition?In this thesis, I tried to answer the above question by using model species: seagrassZostera noltii, the lucinid bivalve Loripes lucinalis and a migratory avian predator, the redknot Calidris canutus, in a intertidal tropical ecosystem of a UNESCO Natural World HeritageSite, Banc d’Arguin in Mauritania (West Africa).BOTTOM-UP FORCES – TROPICAL SEAGRASS MEADOWSSeagrass beds are essential for coastal zones worldwide, as they provide coastal protection,act as carbon and nutrient sinks, and serve as keystone habitat for economically valuablespecies such as fish. In addition, seagrass beds are known as ‘ecosystem engineers’. Suchecosystem engineers are able to alter the physical conditions of their environment whichis not only beneficial for themselves, but for many associated species, thereby also called‘foundation species’. For example, they create a positive feedback by reducing hydrodynamics,stabilizing sediments and accumulating organic matter from the water column.As a consequence however, seagrass beds also create a negative feedback, because accumulationof organic matter stimulates decomposition controlled by sulphate-reducingbacteria that produce sulfide, being highly toxic to seagrasses and all other life. How seagrassbeds cope with sulfide stress remained a mystery for a long time. Our researchshowed that here the three-stage mutualistic interaction plays an important role. Seagrassesengage in a mutualistic interaction with lucinid bivalves (L. lucinalis) and theirsulfide-oxidizing, gill-inhabiting bacteria (‘endosymbionts’) to reduce sulfide stress, resultingin a positive feedback. The gill bacteria consume the toxic sulfide and use it as anenergy source which is also transported to the bivalves. The bivalves and their endosymbiontsnot only profit from sulfide that is indirectly provided by seagrasses due to organicmatter trapping, but also from oxygen released by seagrass roots (chapter 2).SUMMARY172DISRUPTION OF BOTTOM-UP FORCESThere is growing evidence that alteration of the bottom-up forcing can lead to major disturbanceof the ecosystem. For instance, disturbances such as habitat loss and fragmentation,eutrophication, and global warming have led to biodiversity loss. Simultaneously,disturbances change the interaction between bottom-up and top-down forcing. As in ourseagrass system, many other marine foundation species, like salt marches and corals,depend on mutualistic interactions where they additionally reduce physical stress or gainresources. This strong community-wide dependence on these interactions in these ecosystemsalso incurs a clear risk: they bind multiple species to one common fate, and disruptionof the bottom-up forces may therefore lead to habitat degradation and biodiversitylosses. In this thesis, we used a combination of experiments, field surveys, and descriptivedata, computer models and GIS analyses to investigate the breakdown of a facultativethree-stage mutualistic interaction.Analysis of Normalized Difference Vegetation Index (NDVI – a proxy for seagrasscover) suggests a decrease of seagrass in the same period as a drought and heat events.This period was followed by a sharp decline of seagrass. Although both 2010 and 2011 inMauritania were reportedly warm and characterized by drought, our analyses of localclimate data suggest that the summer of 2011 was particularly warm, windy and dryleading to a high evaporative demand. In chapter 3 we investigated whether the die-offcould be related to low-tide desiccation stress caused by drought, wind and heat anomalies,and if disruption of a feedback was an important factor of the observed decline.Before showing changes over time, we performed spatial analyses showing that seagrasscover (NDVI) was lower at higher elevation, suggesting that desiccation stress duringlow-tide exposure reduces seagrass cover at higher elevations. This landscape-scale wideNDVI decrease was supported by ground observations demonstrating a 50% decrease inseagrass cover between 2009 and 2013. Field surveys demonstrated that degrading seagrasspatches had significant lower Loripes densities and higher sediment sulfide levelscompared to healthy patches. Experimental manipulation of Loripes densities confirmedthat the loss of mutualism strength, triggered by desiccation stress, enhances seagrassdegradation (chapter 3). These findings were supported by simulations of a parameterizedmodel, which demonstrated how the mutualism stabilizes intrinsically unstable seagrassbeds by alleviating sulfide toxicity. However, a minor increase in seagrass mortality(as a proxy for desiccation stress) triggers mutualism breakdown, causing seagrass degradation(chapter 4).Ecosystem shifts following gradual environmental change or perturbations of strongpositive feedbacks have been described for a wide range of systems. Several studies haveshown that subtle temperature increases can lead to shifts in ecosystems states. Strongpositive feedbacks can cause alternative states (i.e. bistability), i.e. due to the changingenvironmental conditions a critical threshold is crossed causing a shift to an alternativestate. The feedbacks described in our seagrass ecosystem may indicate that there mightbe alternative stable states (chapter 3). To examine whether our system shows bistabilitySUMMARY173we used potential analysis, a method for detecting feedbacks and alternative states. Ifseagrass changes gradually in response to enhance desiccation stress over the elevationgradient, the frequency distribution of seagrass cover should be unimodal, whereas feedback-driven sudden shifts between two states would typically result in a bimodal distributionfrequency. A potential analysis can identify peaks in the frequency distribution(‘attractors’) over an environmental stress gradient.Although the potential analysis of NDVI data and model simulation results suggestfeedback-driven sudden shifts instead of a gradual response, we did not find evidence foralternative stable states (chapter 3). Instead, the strong mutualistic feedback causes theoccurrence of so-called “slow-fast” cycles when environmental stress is enhanced, – a phenomenonvery similar to what was proposed as a potential explanation for cyclic shifts inshallow lakes. In our case, the system shows abrupt transitions between a seagrass- andbarren state, the states are not stable but tend to stay relatively long in these conditions.Potential analysis on the model results showed that slow-fast dynamics create the samekind of bimodality in frequency distributions as alternative stable states. So far, this kindof multimodality has been attributed to alternative stable states. However, we need to becautious drawing conclusion on potential analysis as there are multiple examples of systemswhich may have slow-fast dynamics. Therefore, it is important to have sound mechanisticinsights of the system when interpreting potential analysis. Obviously, there is aneed for debate and further investigation on this topic to put these finding into perspective(chapter 4).TOP-DOWN EFFECTS OF RED KNOTS?Banc d’Arguin is the main wintering area of the red knot, where the nominate subspeciesC. c. canutus feeds intensively in the intertidal seagrass beds on molluscs. Because, Loripesis the most abundant prey and of high energetic quality, one would expect that Loripes isthe main prey for red knots, potentially creating a strong top-down effect by disruptingthe mutualistic seagrass-lucinid interaction. In contrast to our expectations, faeces analysisrevealed that on average only 50% of the red knot diet consisted out of Loripes (chapter5), and the local depletion (‘giving-up-densities’- GUD) was far less than expectedbased on earlier experiments in the Wadden Sea. Recent studies showed that the lowerdietary contribution of Loripes could be explained by mild toxicity of these bivalves due tothe stored sulfide compounds. Therefore, red knots need to counterbalance their diet withother bivalves.The higher GUDs came to a surprise as red knots have a unique sensory organ in thetip of the bill to detect hard-shelled prey buried in soft wet sediments without direct contact(‘remote touch’). So, a priori one would expect that red knots would be capable ofdepleting Loripes in the soft sediments of the seagrass beds, but on the contrary, ourresearch showed that in high seagrass densities red knots detect their prey by direct touchrather than remotely. Physical modelling of the pressure field build-up around a probingSUMMARY174bill showed that in seagrass the pressure field to no longer reveal the presence of the prey(chapter 6). In fact, seagrass indirectly conceals its detoxifier for its main predator, byobstructing the remote touch, and, in combination with the toxicity of the bivalves, preventingdisruption of the feedback. It appears that a strong trophic cascade interferingwith the mutualistic seagrass-lucinid interaction, which could be expected based on thelarge population size of the avian predator, is lower than expected.In degrading seagrass beds, however, top-down effect may start to play a role again asknots may find their prey more efficiently in this type of habitat. In chapter 6 we arguethat the searching efficiencies of red knots decreased with seagrass biomass, as a consequencered knot should choose feeding patches with no or low seagrass densities andhigh prey densities. However, in chapter 7 we could show that such sites hardly exist inBanc d’Arguin. The mudflats here are either sparsely covered by seagrass with low densitiesof prey that should be relatively easy to find, or densely covered with high densitiesof prey that should be harder to detect. Intake-rate maximizing red knots would have tofind an optimum at intermediate seagrass densities. Indeed, as predicted, our experimentsshowed that red knots attained high searching efficiencies at low prey densities indegrading patches with little or no seagrass, and low searching efficiencies with high preydensities in dense seagrass beds. Using field observations on red knot densities, weshowed that red knots in the field are found at intermediate seagrass densities whereintake rates potentially are maximized. Although we could not differentiate betweenactive choice or random choice. Regardless the mechanism behind their distribution,intake rate were highest on degrading seagrass (chapter 7), therefore, the top-downeffects of red knots may interact with the die-off event by depleting detoxifying Lucinidbivalves and possibly accelerating seagrass degradation.INTERPLAY BETWEEN BOTTOM-UP AND TOP-DOWN EFFECTSTo investigate the top-down effect of red knots the model presented in chapter 5 wasextended, by adding predation pressure by red knots. Here, I assumed an intake rateaccording a functional response type II, Loripes depletion rates based on the toxic-constraintand assumed a negative relationship between searching efficiency and seagrassshoot density, based on our experimental findings in chapter 6. Interestingly, the modelpredicted that red knots may not be able to disrupt the mutualistic feedback in healthyseagrass. One may expect that this is due to the obstruction of seagrass, but in contrastthis was the consequence of the toxic constraint, which limits red knots in their Loripesintake rate (chapter 8). Model bifurcation analysis showed that predation pressure shouldbe extremely high to have an effect on the system. Red knot feeding densities shouldexceed on average 170 birds ha-1 to consume enough Loripes to push the system over acritical threshold to disrupt the mutualistic feedback causing unstable cyclic dynamics(note that average red knot densities in Banc d’Arguin may reach 6–10 ha-1). In degradingseagrass, however, as argued above, red knots may affect the dynamics of a seagrassSUMMARY175bed. A more thorough modelling examination showed that, red knots, in natural densities(10 ind. ha-1), are indeed capable to speed up the degrading event by depleting Loripescausing further accumulation of sulfide and seagrass degradation. Therefore, we mayconclude that the top-down effect of red knots likely accelerates the runaway degradationof the seagrass beds. The modelling exercise in chapter 8 is obviously a simplification ofreality and interpretation of these results should be taken with cautious. But this theoreticalexercise is a step in the direction to better understand the interplay between top-downand bottom-up forces in general, but especially for the Banc d’Arguin intertidal seagrassecosystem.In conclusion, this thesis shows that bottom-up forces are driving the Banc d’Arguinecosystem, although, bottom-up and top-down effects have synergistic impacts on ecosystemfunctioning and that their interaction can explain rapid and large-scale ecosystemcollapse. Although our model species, the red knot is studied in detail, it appears thatprey detection works significantly different in seagrass beds than on bare mudflats in theWadden Sea. This thesis also points out a risk of mutualism dependent marine ecosystems:environmental change can disrupt inherent mutualism-driven positive feedback,causing severe ecosystem change. I therefore emphasize that conservation and restorationshould not focus on a single species or stressor type, but should instead develop a moreprocess-based integrated network approach taking into account trophic and non-trophicinteractions (e.g. mutualism, predation and habitat modification) to be able to predictecosystem response to environmental change. |
