Comparisons between terrestrial and marine ecosystems are generally not in the main stream of scientific literature even though Webb (2012) listed several points for which the transfer of knowledge and concepts related to one or to the other system would benefit our understanding of both. Even sharing this view, the leading hypothesis behind this contribution is that the pelagic system, where the dominant biotic component by number and biomass is microscopic, has specific features which strongly differentiate it from the above-the-surface terrestrial systems. Due to this, climate change, i.e. changes in temperature, precipitation and most importantly in the dynamics of the two fluid media, atmosphere and ocean, act with different mechanisms which prevents proceeding with analogies in many cases. In addition, the non-linearity of most of the processes and responses to perturbations requires, in order to obtain reliable forecasts or hindcasts, a detailed analysis of the path followed by the system which is normally overlooked in the step-change simulations or projections.

terrestrial, marine, climate change impact, LTER

Organisms, communities and ecosystems are continuously exposed to variations of weather components, namely, solar radiation, temperature, atmospheric pressure, humidity, cloud coverage, precipitation, wind speed and direction. These variations may be significant over the short natural cycles of daily and seasonal periods. However, their medians and the squared deviations around the median over a few decades are generally constrained within narrow intervals for each specific geographic location. Those characterise the climate of the specific site (IPCC 2007, IPCC 2013). Therefore climate is just the emergent pattern of weather variations in a specific site over a few decades. If the values, mentioned before, change systematically beyond the typical ranges then, by definition, a climate change is taking place. This raises the question of why climate change, which is a persistent change of the median values of weather components, generally lower than the short term variance of weather variability, has a significant impact on ecosystems.

The weather components listed above are all abiotic processes/variables of the atmosphere or the sun. The reciprocal feedback between abiotic atmospheric processes and biota has been well established in the last decades (IPCC 2013) but, by tradition, we still speak of climate forcing and ecosystem response. The reason may be that three of the key processes that affect the climate system, the radiation emission by the sun, directly and the earth rotation and tectonics, sensu lato, indirectly, are substantially out of the control of biota.

Considering the above definition of climate, another question arises: is a significant climate change presently going on? Recalling the previous statements, climate change means that the weather/climate variables display a persistent change in space and time at regional and global scales. This has always occurred in the past. Primarily because of the orbital coupling of sun-earth system whose main manifestation are the Milankovitch cycles (Berger 1988) or for climate system internal dynamics, such as pluridecadal oscillations (Mann 2007) and also for exceptional events causing dramatic changes in climate leading to drastic reshaping of soil occupation, mass extinctions etc (Whiteside and Grice 2016). Several indicators prove that we are going through a climate change at a global scale (IPCC 2013). The current debate on the issue is focused on whether the present climate change is faster than what could be expected because of ‘natural’ processes, whether this is causing unprecedented shifts on earth ecosystems and, more importantly, whether the anthropogenic activities are causing it. IPCC reports (IPCC 2007, IPCC 2013), as well as numerous studies (Scheffers et al. 2016, amongst others), provide convincing evidence that this is the case for the latter. Therefore, aside from designing prevention and mitigation policies, it becomes relevant to predict how earth ecosystems will respond to this accelerated climate change. This is not an easy task, because of the complexity of natural systems, having a vast number of interactions and non-linear responses that characterise their functioning. This is particularly true for marine ecosystems, which are definitely less known than terrestrial ones. This leads to a further question: do the trends observed in terrestrial systems, which are easier to observe, provide the needed conceptual schemes to predict the response of marine ecosystems to climate change?

Despite the strong interest for climate change impact on earth ecosystems and the fundamental effort carried out by the IPCC, the interaction between ‘terrestrial’ and ‘marine’ scientific communities is not frequent and the number of studies dealing with differences between the two systems is small, if compared with the studies on individual systems.

Literature on climate change and its established or potential impacts, is incredibly vast and it is beyond the scope of this review to summarise it all. The focus of this contribution will be the differences and similarities between terrestrial and marine ecosystems that are relevant in the analysis of the response to climate change. This will aim at highlighting areas of research needing more exploration, hoping to stimulate further discussion on the topic.

While our immediate perception of climate is more related to temperature and rain, all other variables, mentioned in the previous section, contribute to shape the climate. These interact via several feedbacks, involving all of them varying in time. An instantaneous set of values for these or their means and variances, would only provide partial information on how climate might affect us. A typical example is the difference between a mild continuous rain and an intense storm producing a flood, with the same total amount of water having fallen. Therefore, a more appropriate characterisation of climate would be a representation of the contemporary change of all the state variables in a multidimensional space, i.e. a phase space, which would design a trajectory. This trajectory, even if purely phenomenological, would be a much better descriptor of the climate and would be more informative of its potential impact on biota. Likewise, changes in species abundance would not fully characterise the biotic response. Their trajectory mapped in the phase space, which reflects the rates associated with their activities and the consequent interactions, would instead be the best descriptor of how the system functions (Tett et al. 2013, Crise et al. 2015). A prediction of the shape of both trajectories in the time to come is the main scope of climate studies. Due to the non-linearity of the processes in complex systems, as atmosphere and ecosystems, each point on those trajectories depends on the previous path along that trajectory, meaning that simulating changes for a sudden step-variation of state variables might not properly provide information on how the final status of the analysed system will be. We are then confronted with two problems and correspondent knowledge gaps. The first problem is to develop methods for analysing the changes along the trajectories followed by all the state variables linked to the main climate drivers which result from their mutual interactions. A similar process should also be carried out for the biotic component with the added complexity of the reciprocal feedback between biotic and abiotic processes. This approach is adopted with the so-called transient simulations to distinguish them from the equilibrium simulations (e.g. Millar et al. 2015)

The second problem relates to the biotic response which has two components which we can briefly summarise in acclimatisation and adaptation. The first is related to the tolerance of the organism to changes in environmental conditions and may involve internal biochemical adjustments up to epigenetic modifications. The second is instead related to modifications at genomic and genetic levels. While information that relate to the first component can be generated in experimental set-ups or gathered from regular in situ observations, the extent and the characteristics of the second process are unpredictable. What might be within reach in the near future is an estimate of the probability of possible genomic/genetic changes building on an increased knowledge of genome dynamics, the ability to evolve and speciation rates for different class of organisms.

To overcome our knowledge gaps on many mechanisms that drive the time-dependence of the trajectories, the present approaches rely on two strategies: i. they reconstruct the trajectory by numerically simulating a reduced set of processes (e.g. Sokolov et al. 2018); ii. they mine the data with more and more advanced statistical techniques to extrapolate from present information (e.g. Sarhadi et al. 2017). However, both approaches, with their intrinsic limits, focus on the final state of the systems. Even when a reliable prediction is generated, the underlying trajectory is seldom analysed and, therefore, the complex interactions that drive the system to the final state are overlooked. In addition, transient simulations are generally carried out for one or a reduced set of forcings, for example, the progressive increase of CO₂. This implicitly assumes that all changes are caused by a primary driving change, the increase of temperature, which is due to the increase of CO₂. As discussed below, the above limitations affect our predictions of marine ecosystems more than those of terrestrial ecosystem, because of their different functioning.

Those ecosystems trajectories, discussed above, develop in the phase space over time. Time is, therefore, the shared context within which all the interactions and changes take place. Time is not the driver of the changes, which depend on fluxes of matter, energy and information amongst the ecosystem components, but time is the main scaling factor which allows for characterising, quantifying and comparing the changes.

The fundamental question in ecology is how the ecosystems function or, which is substantially the same, which are their dynamics? Tracking the different states in the phase space of the ecosystems is the prerequisite for answering that question. The evolution of Man has been strongly coupled with his capability of exploiting natural resources which, even when Man was a gatherer or a hunter, had to rely on associations but also on predictions. The birth of agriculture is based on having acquired knowledge on the coupled cycles of plants and environmental forcing, i.e. on parts of ecosystem dynamics. All this leads to the essential role of ecological time series. All the present knowledge about Earth functioning is based on our reconstruction of its dynamics for the past and for the present, made possible by observations over time. Long Term Ecological Research (LTER), a very recent formalisation of a long implicitly known practice, is often perceived as a specific niche of ecological research. In fact, the only difference with any other ecological observations is the time scale, by definition longer than episodic observations focused on specific processes and, often, the sampling in a fixed, spatially limited area. Both traits often have originated criticisms, for the long term sustainability, the former and for their representative nature, the latter. The former is more dependent on societal awareness and willing-to-support, therefore asking for an improved outreach activity. The latter, instead, can be overcome by promoting observational strategies that integrate the periodical observations with sampling efforts focused on the characterisation of the spatial context and/or on specific processes that could be revealed by periodic sampling. This can be better achieved by building networks of sites and is fundamental for directing ecological research (e.g. Vanderbilt and Gaiser 2017, Zingone et al. 2019). The discussion that follows is largely based on the information collected in observational efforts over time.

Besides the obvious visible differences between the terrestrial and marine environments, e.g. in the first animals walk and in the second they swim, it is still an object of discussion if all the other differences prevent the identification of general ecological rules, valid for both realms. The existence of general rules in ecology, which would allow prediction of the structure of any ecosystem, once assigned initial conditions and fluxes, is part of a long lasting debate. Lawton (1999) argued against the existence of such laws, even if he admitted that macroecological patterns may display regularities beyond the ensemble of contingent events that may involve the different organisms in the ecosystem. This would imply that the final state reached after a perturbation or a transition would not be strictly dependent on the trajectory followed by the system, at least at a coarse, macro scale. This, in turn, might imply that some general structural properties, e.g. architecture of food webs or mechanisms, e.g. trophic or mutualistic interactions, should be shared by all ecosystems, independently from the components, which would be more exposed to contingent events. From this general statement, one could then assume that marine and terrestrial systems should share several common traits and that the separation by the two scientific communities is more related to tradition or differences in observation tools, than to fundamental differences in the systems (e.g. Webb 2012). However, the analysis conducted by Chase (2000) on various characteristics of the food webs in the two environments, amongst which the existence of cascades, the number of trophic levels and the dominant size of primary producers, led him to conclude that some differences, at least in the functioning and structure, exist. This would suggest that their response to climate change might also follow different paths, a possibility also discussed by him (Chase 2000).

Indeed, there are several evident differences between the two systems, especially if one compares the pelagic systems with the above-surface terrestrial ecosystems which are the most distant in terms of characteristics and are the focus of this contribution.

First of all, let us consider the dimensionality. Pelagic marine systems are fully three-dimensional and the total volume is occupied by organisms. In terrestrial systems, the third dimension has a limited thickness, with the exception of the overlying atmosphere which, however, hosts only a very small fraction of biomass (Bar-On et al. 2018). For marine organisms, this requires a significant tolerance to an extended pressure range which is another specific trait of the marine systems. On the other hand, on land, differences in altitude may determine drastic changes in environmental conditions on short linear distances and, therefore, much sharper gradients. Another remarkable difference is the reduced gravity in the marine systems with a parallel increase in friction during movement. Energy and signal transfers are also substantially different. Light is rapidly absorbed by water and modified in its spectral properties (Kirk 2011), implying less available radiant energy and potential constraints. Light is also attenuated in the terrestrial environments but, above the surface, mostly under canopies and with a different pattern of spectral modulation (Depauw et al. 2012) On the other hand, spectral changes with depth can be a source of detectable signals by organisms (Jaubert et al. 2017). This is partially compensated by acoustic signals which propagate faster and over longer distances in the water (Hovem 2010), a fact which explains why sound and mechanical signals are widespread in marine ecosystems (Dusenbery 1992). However, there are two key differences between the two systems. One is the water availability. Water is the fundamental element of life, at least the life that we know. Terrestrial organisms rely on water supplied by precipitation or humidity in the air and in the soil while marine organisms are embedded in water, even if this means that they have to handle the osmotic pressure of the medium. The other is the medium on which organisms rely to make their life. Terrestrial systems are organised on a fixed substrate and the mobile component, the atmosphere, is where the organisms just transit to move from one fixed site to another. In the marine environment, the medium is mobile with most of the biomass living in it (Bar-On et al. 2018) and in a size range such as to make it prevalently transported by the movement.

Those differences are significant, but not all of them may directly affect the response to climate change. Some of them are crucial, i.e. temperature, water availability and medium motion, others enter the game indirectly, e.g. dimensionality, while those remaining have relevance from an evolutionary point of view but do not modify the impact of climate change on the two systems.

In Table 1, the three climate components, mentioned above, are listed together with the direct dominant effect that a change in their amplitude would have on terrestrial and pelagic ecosystems. In the last row, there is an additional component, the movement of the fluid media (atmosphere and oceanic water) which is strongly intertwined with the others, in fact it depends on them, but exerts a complex, direct, very specific impact on biota.

So far, I have discussed the different mechanisms by which climate acts on the marine pelagic and above-the-surface terrestrial ecosystems. However, between the two, there are no rigid barriers and one may wonder which could be, or could have been, the reciprocal feedbacks within the two sub-systems under climate forcing and the impact on biota. Indeed, if we exclude the few organisms that divide their life between solid Earth and ocean, such as, for example, some birds, a few mammals etc., the reciprocal feedback between the two systems, even when driven by biota, acts via abiotic players, i.e. the atmosphere and the water. Though, it can act on a wide range of scales up to the macroevolutionary timescale, thus being coupled with, but not necessarily driven by, climate. The iconic example is the origin of oxygenic photosynthesis which, via the accumulation of oxygen produced essentially in the marine environment (Sánchez-Baracaldo 2015, Fischer et al. 2016), also caused a drastic change in terrestrial environment and biota. More recent events, on a geological time scale, are the sapropel depositions, which are accumulations of reduced carbon in the sediments of the Mediterranean Sea (Rohling et al. 2015). Those events are driven by co-occurring processes, such as Monsoon intensification, increase in precipitation and runoff. There are several lines of evidence that the consequent increase of stratification and, possibly, the increase in export production fuelled higher nutrient inputs from land, with a parallel weakening of the thermohaline circulation in the basin and caused a decrease in oxygenation rates of the deep layer, leading to euxinia or anoxia. Any change in the hydrological cycle on land which, as discussed above, is also affected by terrestrial plants and activity, impacts, via runoff, the marginal seas and coastal areas (Meybeck et al. 2006) and can track climate fluctuations via the paleobiological changes and their isotopic signatures (e.g. Sprovieri et al. 2012). A third example is the observed increase in iron transport from land during glacial times initially hypothesised by Martin (1990). The harsher climate on land during those periods eventually produced an increase in iron transport to the ocean with a parallel increase in carbon vertical export and, very likely, a change in community structure. The impact of the recorded increase in iron transport to the ocean is still at the centre of an intense debate, but the evidence supporting the coupling between changes induced on terrestrial environment on land and those in the marine ecosystem is definitely robust. A very crude generalisation would be to separate the scales. At the interface between land and sea, i.e. on a small scale, climate change on land would act via runs-off and would be more frequent from the land to the sea, even though it can be hypothesised that habitat destruction on the near-shore due to storm intensification might impact on on-shore habitats. On a larger scale, the atmosphere would likely play a prominent role, also via global biogeochemistry.

We discussed above that temperature changes play a role at multiple levels. Prediction of temperature future trends is still a great challenge for Earth system models (Bonan and Doney 2018) due also to the feedback of biota. This in turn depends on their physiological fitness, as well as on their capabilities for acclimatisation and, eventually, modification of their fitness via mutation. The importance of biota feedback can be generalised to many more processes and is the key challenge that the scientific community will face in the near future. This can only be obtained via a holistic approach that should merge classical observational methods with the recently developed new tools, such as bio-optics, bio-acoustics and -omics (Karsenti et al. 2011, Coles et al. 2017, Crise et al. 2018, amongst others). It should consider in more detail the multiple interactions occurring within communities which are differently modulated in the marine and terrestrial environments and display plasticity and complex outcomes (Tylianakis et al. 2008, D’Alelio et al. 2016, amongst others). It should be continued in time as the LTER community is showing for many years. In fact, most of the issues discussed in this paper were stimulated by the observations conducted over the last decades at the LTER station MC in the bay of Napoli (Zingone et al. 2019). This information, especially if integrated with modern techniques and approaches, especially those based on bioptical and molecular methods, can feed a new generation of models (Coles et al. 2017, Stec et al. 2017, D’Alelio et al. 2019).

We are facing a faster change in the environment even with regards to climate because of anthropogenic actions. We know from the past that climate change induces ecosystem changes. However, our data result from a natural filtering process that transmits only the prominent changes. The importance of ‘history’ has been stressed before. In the background section, I raised the question on why biota respond to changes in environmental conditions that are often smaller than the variations they experience in one single year. I believe that this too is related to the time course of the change. Short term fluctuations are tolerated by organisms which, in addition, tune their life cycles to the most suited conditions for them, meaning that many of them do not experience the whole range of variations of the seasonal cycle or adopt different solutions to cope with it. Besides the cycle of illumination which is the same for the ocean and the land, terrestrial ecosystems experience a more fluctuating environment with wider ranges of variations for the key climatic components. While the heat capacity of seawater buffers the temperature, an equivalent buffering role is partially played by the soil with respect to water availability. On the other hand, the real modulating player is the motion of the medium, on land because of the impact of wind in moisture transport and evaporation and evapotranspiration and in the ocean because of the currents. The main difference is that, on land, the inertia is much higher because the motion of medium changes the conditions but does not move the organisms, whereas in the ocean, organisms in many cases move with the water. Nectonic organisms are in-between because they may move independently from the medium, to some extent. This does not always mean that they can escape the impact of changes, because a moving environment also changes their environment.

In any case, the conceptual framework for the two systems has some relevant differences and even if one wants to build on unifying theories, the real impact must be based on the integral of the changes over time which depends on different processes on land and in ocean.