1 Introduction

Resilience (from Latin resilire, to leap back) is the ability of the ecosystem to recuperate its structure and functions after a perturbation. In the ecological literature, the term resilience has been used inconsistently over the years (Standish et al. 2014), perhaps depending on how ecosystem dynamics and the role of natural and anthropogenic disturbances are interpreted (Nyström et al. 2000). For example, it has been equated to recovery or extended to embrace the sister concept of resistance (Côté and Darling 2010). The latter term (from Latin resistere, to stand back) should rather refer to the opposition that the ecosystem offers to a perturbation. Clearly, resilience and resistance are the two sides of the same coin: the stability of the ecosystem in the face of environmental change (Montefalcone et al. 2011). This chapter focuses on resilience, measured as the time needed to recovering the preexisting conditions after a perturbation. Resilient ecosystems recover fast; non-resilient ecosystems recover slowly and may be prone to hysteresis (from Greek husterein, to be behind), i.e., the lagging of the effect when the cause has ceased, or even to phase shift (from Greek phasis, appearance, and Old English sciftan, to arrange), i.e., a substantial modification of the structural state of the ecosystem.

Marine ecosystems are globally exposed to a number of pressures, such as increased nutrient and sediment load from land, littering and pollution, overfishing or other direct human impacts, extreme weather episodes and thermal anomalies, pathogen diseases, and alien species invasion (Rossi 2011). These pressures are producing biodiversity loss and may lead to dramatic change in ecosystem functioning and in the provision of services. Assessing resilience of marine ecosystems has therefore become a prime concern for ecology and conservation (Hughes et al. 2005; Lotze et al. 2011) and research in the field has grown tremendously in the last few decades (Duarte et al. 2015).

Among marine ecosystems, animal forests are normally shaped by long-lived organisms (mostly anthozoans and sponges), which can subsist for centuries or even millennia (Schuhmacher et al. 2002; McMurray et al. 2008; Parrish and Roark 2009; Bo et al. 2015). In the sessile benthos, longevity typically correlates with scarce reproductive potential and poor dispersal capacity, and the resilience of animal forests after episodes of mass mortality of these K-selected species is therefore expectedly low. Recovery may take long, increasing the risk of hysteresis and phase shift (Rossi 2013). While information about mass mortality episodes has accumulated in the last decades, our knowledge about patterns and timing of recovery is still inadequate. Recovery (from Old French recovree, past participle stem of recovrer, in turn from Medieval Latin recuperare, to regain health or strength) may be defined as a return to a normal state, but such a “normal state” is often not known for the marine animal forests. Estimating magnitude, patterns, and trajectories of change is indeed necessary to evaluate the possibility of recovery of animal forests but requires information on their previous conditions. Unfortunately, this kind of historical data on marine animal forests are extremely rare for most regions of the world ocean due to the challenging difficulty of the study of deep- and cold-water habitats, such as seamounts, cold-water coral reefs, the Mediterranean coralligenous, etc. (Schlacher et al. 2010; Gatti et al. 2015a). The nature, tempo, and trajectories of ecological changes that follow a major disturbance are therefore virtually unknown for this kind of animal forests (Williams et al. 2010).

Tropical coral reefs , thriving in warm and comparatively shallow waters, are easily studied and undoubtedly the best known example of animal forest (Rossi 2013). They may therefore provide ideas and concepts to be applied to the study of other animal forests. Based on literature analysis, coral reef recovery after major disturbances (bleaching, hurricanes, crown-of-thorn attacks, diseases, etc.) may require several years to decades (Nyström and Folke 2001; Bianchi et al. 2003; and references therein). Such different estimates could depend on geographic location, reef type, exposure, and depth but also on how resilience is actually measured. Most useful descriptors and metrics include either demographic or community parameters: recruitment, mean colony size, species diversity, similarity with the premortality condition, percent coral cover, and reef structure complexity (Bianchi et al. 2006). In addition, the threshold of the specific disturbance considered is central to measuring resilience (Standish et al. 2014), as the same animal forest may be highly resilient to minor disturbances but not resilient at all to major ones.

A major example of a global disturbance that hit a paradigmatic animal forest has been the mass bleaching event that followed the extreme El Niño – Southern Oscillation (ENSO) episode of 1998 (West and Salm 2003; Obura 2005). The Maldives have been among the most affected countries, with 60–100 % coral mortality reported (Bianchi et al. 2003). This chapter uses historical data series on Maldivian coral reefs (Morri et al. 2015), aiming at illustrating their recovery trajectory and taking into account multiple descriptors: (1) recruitment, (2) colony density and size, (3) colony health, (4) cover, (5) diversity, (6) community composition, (7) trophic organization, (8) structural complexity, (9) constructional capacity, and (10) seascape. The Maldivian example will be discussed in order to draw attention to the implications of the choice and use of a specific descriptor to measure the resilience of animal forests .

2 Recruitment

Together with regrowth and asexual reproduction of the survivors, the main step toward recovery of an animal forest following a mass mortality event is obviously recruitment . Recruits may come from local survivors or from distant parents, which may be either brooders or broadcasters . Brooders incubate embryos on their own body, and larvae are released when already competent for settling. This normally implies short pelagic life and hence limited larval dispersal capacity. In broadcasters, fertilization normally occurs in the water, and larvae can live free for a comparatively long time and thus disperse over longer distances. Early phases of recovery may therefore be characterized by the recruits of the same species that previously constituted the forest if the local survivors are brooders, otherwise may be more prone to the colonization by different species. However, this somewhat simplistic schematization may be complicated by differential dispersal capacities and patterns of connectivity within and among reefs. Larvae from brooders may travel long distance when attached to floating object (Jokiel 1992), while long-lived larvae from broadcast spawners may stay around their natal reef if local currents prevent their large-scale dispersal (Sammarco and Andrews 1988; Black et al. 1991).

In the Maldives, hard coral recruitment was quantified from 1999 to 2014 by counting colonies smaller than 5 cm in diameter within replicated quadrats of 1 m2, randomly placed in the reef flat and upper slope. Recruits were divided into three categories: Acropora, Pocillopora, and “others” (Cardini et al. 2012); especially in the case of “others”, which included massive and encrusting species, particular care was taken in the field to avoid counting surviving patches of preexisting adult colonies. Despite high spatial variability, a general temporal trend was recognizable (Morri et al. 2015), and four distinct phases could be described: (i) 1999, (ii) 2000–2003, (iii) 2004–2008, and (iv) 2009–2014 (Fig. 1).

Fig. 1
figure 1

Time trends between 1999 and 2014 of the number of coral recruits (mean ± standard error) belonging to three categories (Acropora, Pocillopora, and others) in the coral reefs of the Maldives (Data from Morri et al. 2015)

Full size image

In 1999, many recruits were found on dead coral colonies, especially in shallow reefs. The most abundant among these recruits were encrusting Agariciidae , followed by massive Faviidae and Poritidae , all belonging to the category “others”. Both encrusting and massive corals suffered comparatively lower mortality rates, and the high recruitment observed nearly one year after the bleaching has been interpreted as the result of an “emergency spawning ” following the thermal stress. Emergency spawning could be a paradoxical strategy to face environmental change and can be compared to hormesis , the phenomenon for which a stress factor produces positive effects at a sublethal level (Bianchi et al. 2003). Recruitment by Acropora, which suffered the highest mortality in 1998 (Bianchi et al. 2006), was extremely low in 1999. Although scarce, some Acropora colonies reached 14 cm in height. This size is compatible with the yearly growth rate reported for several species in the genus, and it can therefore be inferred that the first recruitment waves for this coral arrived soon after the mortality event, possibly from colonies surviving in deep refuges. No recruits of Pocillopora, which also suffered high mortality in 1998, were recorded in 1999. Between 2000 and 2003, recruitment of all corals was extremely low, possibly due to the scarcity of living adult colonies. From 2004 to 2008, recruitment rate increased, especially for the category “others”, which resulted dominant. Adult colonies of Acropora and Pocillopora had been nearly extirpated in 1998, and the recruitment of these corals could start only after the newly settled colonies matured. Finally, in the period 2009–2014, recruitment rate by the others slowed down while that of Acropora increased. Acropora became codominant with the “others”; Pocillopora recruits, always scarce in the previous phases, became comparatively abundant, especially on shallow ocean reefs.

Differential success of the three coral categories was due to a combination of recruitment and post-settlement survival, while regrowth and asexual reproduction (such as fragmentation) of older colonies appeared unimportant (Morri et al. 2015). Recruit community composition in 2014 roughly mirrored adult community composition: acroporids, massive and encrusting corals (i.e., the “others” in the recruitment counts), and branching corals (mostly pocilloporids) characterized Maldivian coral reefs as in 1997–1998.

3 Colony Density and Size

The canopy of animal forests is defined and shaped by colonial invertebrates (e.g., cnidarians; in the case of sponges the term colony may be inappropriate). Measuring colony density and size after a major disturbance is therefore of prime importance to evaluate the recovery potential of animal forests.

In tropical coral reefs, the canopy is often formed by tall arborescent species, such as many Acropora corals. In the Maldives, large tabular Acropora corals used to dominate at shallow depths before 1998 (Morri et al. 1995; Bianchi et al. 1997) but nearly disappeared following bleaching. In 2004, Maldivian reefs appeared rich in tabular Acropora colonies again, especially at some sites (Fig. 2). Acropora tables were counted, and their diameter measured, in replicated belt transects (20 m × 2 m) randomly laid in reef flats and slopes. Only colonies with diameter equal or greater than 15 cm were taken into account, as tables bigger than this size are known to stabilize their radial extension at a nearly constant rate of 10 cm · a−1 (Lasagna et al. 2010a).

Fig. 2
figure 2

Examples of Maldivian shallow reef coral communities dominated by large tabular Acropora before (1997) and after (2004) the mass bleaching and mortality of 1998 (Modified from Lasagna et al. 2010a)

Full size image

A total of 192 colonies of Acropora tables was counted in 2160 m2, averaging 4.8 colonies per transect (40 m−2) but reaching 8–10 colonies per transect in quiet lagoon reefs. Most colonies had diameters smaller than 40 cm, but some were larger, up to 105 cm. Assuming a radial extension rate of 10 cm · a−1, colonies larger than 65 cm can be interpreted as the survivors of the mass mortality of 1998; conversely, they may represent the result of enhanced growth rates in uncrowded situations as those characterizing the coral reefs of the Maldives in 2004. Tabular Acropora colonies are known to grow rapidly and to be able to outcompete the neighboring corals – two features that facilitate their recovery after a disturbance.

4 Colony Health

The mortality caused by a disturbance may be total or partial. The former involves the death of the whole colony, and recovery would therefore require recruitment of new organisms coming from outside in the form of larvae issued from sexual reproduction. The latter produces more or less wide necrotic areas in the colony, which remains anyway alive. Recovery, therefore, might in theory take place – at least in part – thanks to the regrowth of living tissue from the borders of necrotic areas. In coral reefs, for example, branching coral colonies, such as Acropora and Pocillopora, typically undergo total mortality, while massive coral colonies (e.g., Faviidae) often exhibit partial mortality (Bianchi et al. 2003).

Quantifying the proportion of colonies totally or partially dead and the amount of damage on the latter ones may therefore be useful to assess the state of the animal forest after a disturbance. Similarly, the repetition of this quantification over time may help evaluating recovery: the persistence of damaged colonies may suggest hysteresis, while a proportional increase of healthy corals will be an unequivocal indication of recovery.

Lasagna et al. (2014) devised a Coral Condition Index (CCI) based on the proportional abundance of coral colonies belonging to six ordinal categories that represent their condition: (i) recently dead, (ii) bleached, (iii) smothered, (iv) upturned, (v) broken, and (vi) healthy. These categories were assigned a different degree of health, according to the following scale: Degree 0 – recently dead corals, whose colonies evidently had no possibility of recovery; Degree 1 – bleached coral colonies, which could survive bleaching if intensity and duration of the causal agent (i.e., high temperature) do not exceed certain limits, beyond which corals die; Degree 2 – smothered coral colonies, which can survive provided that the loose sediment deposited on the branches is removed by physical agents or shed along with coral mucus; Degree 3 – upturned coral colonies, which could express potential for reattachment and growth, especially if landed on solid substrates; Degree 4 – broken coral colonies, which could show signs of growth of new tips and had no evidence of disease; Degree 5 – healthy coral colonies.

The Coral Condition Index (CCI) was computed on the basis on the above hierarchical scale and the abundance of coral colonies, according to the formula:

$$ \mathrm{C}\mathrm{C}\mathrm{I}=\left[\left(\mathrm{H}/\mathrm{T}\times 5\right)+\left(\mathrm{B}/\mathrm{T}\times 4\right)+\left(\mathrm{U}/\mathrm{T}\times 3\right)+\left(\mathrm{S}/\mathrm{T}\times 2\right)+\left(\mathrm{W}/\mathrm{T}\right)\right]\times {5}^{-1} $$

where H is the number of healthy coral colonies, B is the number of broken coral colonies, U is the number of upturned coral colonies, S is the number of smothered coral colonies, W is the number of bleached coral colonies, and T is the total number of coral colonies. CCI ranges from 0 (100 % dead corals) to 1 (100 % healthy corals).

The CCI was applied in the Maldives from 2005 to 2010 to tabular Acropora and Pocillopora colonies, using replicated belt transects (20 m × 1 m) at several depths and across different habitats (Lasagna et al. 2014). The majority of coral colonies were found in healthy conditions, and damaged corals represented on average only 20 % of the total: consistently, CCI ranged from 0.23 to 1.00, with an overall average of 0.84. In general, bleached and recently dead colonies represented the highest proportion of damaged corals. Smothered corals were never abundant and so were upturned corals. Broken corals were most represented in 2005, but their proportion decreased afterward; it should be recalled that the Maldives were hit by the Sumatra-Andaman tsunami of 26 December 2004.

The CCI condenses, integrates, and summarizes information on the status of individual colonies, providing an instrument for the early detection of environmental alteration. It did not show correlated to the number of total coral colonies, which means that it could be successfully employed independently of the coral abundance in the specific site under study. Further, it has the advantage of being of simple use and then applicable in long-term and large-scale monitoring plans.

A high proportion of corals with damage of the highest degree of severity (i.e., bleached and recently dead colonies) leads to a low value of CCI and suggests large-scale disturbances (e.g., climate anomalies). In contrast, moderate values of CCI suggest specific disturbances acting on a small scale, therefore predicting change that can be averted by local management actions.

Co-occurrence of different damage types, resulting from multiple pressures, could produce an increase of instability and fragility of the entire community. The abundance of severely damaged corals (low value of CCI) is indicative of bad reef conditions and may represent a point beyond which, in the worst hypothesis, resilience is eroded, and the consequent hysteresis may lead to phase shift, which will require huge and perhaps hopeless efforts for restoration (Montefalcone et al. 2011). CCI may be proposed as a complementary tool to measure ecosystem health and as such may serve to evaluate the feasibility of restoration plans. As CCI asks for detailing the kind of damage, it could be employed not only for the evaluation of present condition but may also be used as a predictive approach to assess community resilience and to help well targeted restoration programs. The experience in the Maldives by Lasagna et al. (2014) can be exported with no or little change to other Indo-West Pacific localities (Ferrigno et al. 2016), where species of Acropora and/or Pocillopora are normally abundant. Damage due to a specific disease can be added in the formula in regions or situations where diseases are important (e.g., the wide Caribbean). Appropriate taxa, belonging not only to scleractinians but also to gorgonians, antipatharians, sponges, or other canopy organisms, should be individuated in other areas and/or in other kinds of animal forest. Gatti et al. (2012) used the percentage of epibiosis and/or necrosis to assess the health state of the canopy-forming organisms in the Mediterranean coralligenous. The selection of few representative organisms able to show specific responses as a consequence of environmental stresses is a major concern for any index of biotic integrity.

5 Cover

Cover is one of the most universally used metrics to quantify sessile benthic organisms and communities on hard substrate, and is usually expressed as percentage (Bianchi et al. 2004). The mortality induced by natural or anthropogenic disturbances always implies a more or less severe reduction of biotic cover, which is therefore a useful tool for monitoring and assessing both algal and animal forests. In coral reefs, in particular, it is customary to measure the percentage of hard coral cover.

In the Maldives, Morri et al. (2015) collated historical data on hard coral cover obtained from published sources, including grey literature, and their own field surveys. An overall yearly mean of hard coral cover was computed irrespective of reef type, site, and depth zone (flat and upper slope) to describe the general trend of coral cover on Maldivian reefs during the last decades (Fig. 3).

Fig. 3
figure 3

Time trend (mean ± standard error) of hard coral cover in the Maldives, 1956–2014 (Redrawn and modified from Morri et al. 2015)

Full size image

The earliest data on coral cover in the Maldives come from the pioneer work done during the celebrated Xarifa expedition in the late 1950s (Wallace and Zahir 2007). Maldivian coral reefs at that time might be considered near pristine conditions, and coral cover averaged 65–75 %. Early development, from coastal works at the British military base of Addu in the 1960s to increased extraction of coral rock as building material and growing tourism through the 1980s, was accompanied by a fall of coral cover to about 50 %. The pressure regime changed in the 1990s. Maldivian coral reefs experienced not only increased human pressure, implying sediment loading and eutrophication, but also “minor” bleaching events and attacks by the crown-of-thorn starfish Acanthaster planci (Morri et al. 2010, 2015). Local reductions of coral cover down to 10–30 % were nevertheless followed by rapid recovery to values >50 %, suggesting an intrinsic high scope for resilience of Maldivian coral reefs. Things showed completely different after the mass bleaching of 1998, which caused living coral cover to fall to 2–8 % (Bianchi et al. 2003). Although recovery started soon (Bianchi et al. 2006), coral cover has been increasing at an extremely low pace (Lasagna et al. 2008) and returned to values >50 % only in 2014 (Morri et al. 2015). This contrasts with the comparatively fast coral recovery observed in the 1990s after minor disturbances. However, the Maldives of those years were still little impacted by human pressures, while today they are experiencing a tumultuous increase of tourism, urbanization, and coastal development (Nepote et al. 2016).

6 Diversity

The importance of biodiversity for ecosystem functioning has been underlined in thousands of articles and books, and the measure of diversity is often taken as a proxy for the appraisal of ecosystem complexity. The high number of taxa living associated to animal forests make them biodiversity hot spots in the sea, and species richness is one of the recommended parameters to estimate ecosystem recovery after major disturbances (Bianchi et al. 2006).

The first attempt to dress a complete list of the coral species living in the Maldives was done by Pillai and Scheer (1976), largely based on the results of the Xarifa expedition (Wallace and Zahir 2007). Other earlier inventories included the works of Wells and Spencer-Davies (1966) for Addu, of Pillai (1971) for Minicoy (administratively attached to the Laccadives), and of Ciarapica and Passeri (1993) for Felidu and North Malé. Amending previous coral inventories, both Pillai (1972) and Sheppard (1987) recorded 166 species for the Maldives. Unfortunately, no inventory was undertaken in the Maldives immediately after the mass bleaching of May 1998, which might have caused the local extinction of the most vulnerable species. In 2002, Bianchi et al. (2003) recorded 147 species. On the basis of a species-effort curve, they extrapolated a total number of coral species equal to 168, extraordinarily close to Pillai’s and Sheppard’s estimates before the bleaching. Thus, 4 years had been enough to reestablish the preexisting species richness of Maldivian coral communities. Later thorough inventories recorded 180 species from South Malé, Felidhu, Wataru, and Ari (Pichon and Benzoni 2007) and 177 species from Baa (Bigot and Amir 2012) – i.e., figures higher than before the bleaching (Fig. 4). According to various compilations and estimates, including Veron (2000), the coral species richness in the Maldives might well exceed 200.

Fig. 4
figure 4

Time trend of coral species richness in the Maldives (Data from: A, Wells and Spencer-Davies 1966; B, Pillai 1971; C, Pillai 1972; D, Pillai and Scheer 1976; E, Sheppard 1987; F, Ciarapica and Passeri 1993; G, Veron 2000; H, Bianchi et al. 2003; I, Pichon and Benzoni 2007; and J, Bigot and Amir 2012)

Full size image

Notwithstanding present coral species richness is apparently higher than in the past, there are examples of taxa that have virtually disappeared from the Maldives, such as the scleractinians Stylophora and Seriatopora (Bigot and Amir 2012) and the hydrocoral Millepora (Gravier-Bonnet and Bourmaud 2012). A severe reduction in zooxanthellate soft corals has also occurred after the mass mortality of 1998 (Fig. 5), but no information is available about the species concerned by this reduction.

Fig. 5
figure 5

Time trend of percent cover (mean ± standard error) of different benthic organism categories and non-biotic attributes in the coral reefs of the Maldives. Note different scales on Y axis (Data from Morri et al. 2015)

Full size image

In general, knowledge of most marine phyla in the Maldives remains poor and with little background information (Andréfouët 2012), hampering comparison with the pre-disturbance situation. This is equally true for most animal forests, and especially for the deepwater ones, which only recently have been studied with some detail (Schlacher et al. 2010), thanks to the development of emerging technologies such as technical diving and remotely operated vehicles (ROV).

7 Community Composition

The quali-quantitative composition of the communities inhabiting the animal forests may be expected to change after the occurrence of a perturbation. Different species exhibit different sensitivity to a particular disturbance, and the resulting mortality rates can vary greatly among individual species. Early phases of community recovery may experience an increased proportion of r-selected species, which take advantage from the newly available space; on the contrary, recruitment of K-selected species may be delayed. According to the level of hysteresis (Montefalcone et al. 2011), the change in community composition may persist for some time or even result in a phase shift, with a stable community completely different from the one existing before the disturbance. Monitoring the quali-quantitative composition (e.g., the cover of the individual component species) is therefore important to evaluate the stability of the whole animal forest community.

While research on community composition is deemed necessary to understand the disturbance history of coral reefs (Graham et al. 2014), the knowledge of the taxonomic composition of coral reef communities in the Maldives is inhomogeneous and therefore not sufficient for many major groups (Jimenez et al. 2012). However, earlier studies evaluated the cover of non-taxonomic benthic categories, i.e., lumped levels of classification of sessile organisms, combining higher taxa with growth forms (Morri et al. 2010). Using 20 such non-taxonomic categories (Acropora branching, Acropora digitate, Acropora tabular, coral branching, coral foliose, coral massive, coral encrusting, Tubastrea micranthus, Heliopora coerulea, Millepora, corallimorpharians, Palythoa, soft corals zooxanthellate, soft corals azooxanthellate, whip- and wire corals, seafans, algae, sponges, Tridacna, tunicates) plus three non-biotic attributes (sand, rubble, rock), the coral reef communities have been monitored between 1997 and 2014, i.e., before, during, and after the mass bleaching and mortality of 1998 (Morri et al. 2015). This time series offers a unique opportunity to estimate the magnitude of change and the patterns and trajectories of recovery of an animal forest community after a major disturbance.

Each year between 1997 and 2014 (but in 2011), the percent substratum cover of the abovementioned benthic categories was estimated by replicated point intercept transects 20 m long laid parallel to the reef edge in the outer flat and upper slope. Correspondence analysis of the data provided an ordination plot in which the trajectory connecting year centroids illustrates a more or less conspicuous change between 1997–1998 and 1999–2000 and then a slow and irregular trend of return toward the initial situation of 1997–1998 (Fig. 6). Combining this trajectory with the time trend in the cover of the individual category (Fig. 5), five main phases can be recognized: (i) 1997–1998, (ii) 1999–2002, (iii) 2003–2004, (iv) 2005–2010, and (v) 2012–2014.

Fig. 6
figure 6

Ordination plot on the first two axes from the correspondence analysis of a matrix containing percent cover data of 23 benthic categories in 1224 transects in Maldivian coral reefs: first axis (horizontal) explains 15.4 % of the total variance; second axis (vertical) explains 14.0 % of the total variance; the cross depicts the center of the axes. For the sake of simplicity, only year centroids are indicated, together with the average time trajectory between years (Data from Morri et al. 2015)

Full size image

In 1997–1998, before the mass bleaching event, Maldivian reefs were dominated by hard and soft corals, followed by algae and sponges; many other organism groups were also present, while non-biotic attributes exhibited comparatively low cover values. In 1999–2002, following the bleaching, hard and soft corals virtually disappeared, with the exception of massive and encrusting corals, which had been less affected by mass mortality (Bianchi et al. 2003). The cover of other organisms did not change significantly, while that of non-biotic attributes, and especially of rock, increased tremendously. In 2003–2004, the cover of all categories of hard corals started to re-increase and that of rock to decrease; corallimorpharians showed abundant for the first time. In 2005–2010 corals, and especially tabular and branching Acropora and branching corals, continued increasing, corallimorpharians disappeared, and non-biotic attributes remained similarly abundant as in the previous period, suggesting that Maldivian reefs were still in a recolonization phase (Lasagna et al. 2008). Finally, in 2012–2014, although non-biotic attributes showed still comparatively abundant, the cover of hard coral categories returned to the pre-bleaching value. Millepora, however, was never found again, and zooxanthellate soft corals remained scarcer than before the bleaching. All the remaining categories changed little through the whole period of study (Fig. 5).

Throughout this time series, the shift from a coral-dominated state to an algal-dominated state, which often characterizes the global degradation of coral reefs (Cheal et al. 2010), has never been observed. Other studies in the Indo-Pacific demonstrated that grazing by herbivorous fish plays a fundamental role in hampering algal dominance (Hughes et al. 2007a, b). Data on fish assemblages in the Maldives for the study period are not available, but between 2003 and 2005 we noticed unusually large schools of Scaridae and Acanthuridae grazing during our field work; in 2003, we observed abundant Bulbometopon muricatum, an otherwise rare sight in the Maldives.

8 Trophic Organization

The foundation species of animal forests are by definition heterotrophic; virtually all of them are passive suspension feeders or active filter feeders , taking particulate matter (plankton and/or detritus) from the surrounding water. However, in well-lit and even in mesophotic animal forests, autotrophy is still possible. Macroalgae, even if usually not dominant, may thrive more or less copiously amidst gorgonians and sponges, while many coral species, obviously dominant in coral reefs but also present in other habitats, harbor endosymbiotic microalgae belonging to the dinoflagellates and commonly called zooxanthellae . Zooxanthellae are not restricted to cnidarians but also occur in clionaid sponges, tridacnid bivalves, and other animals. Many sponges and some didemnid ascidians harbor zoocyanellae (cyanobacteria), and some sea anemones harbor zoochlorellae (chlorophytes).

These animals hosting photosynthetic endosymbionts challenge the distinction between autotrophy and heterotrophy and well deserve the epithet “zoophytes ” that nineteenth-century naturalists granted them with. They get part of their energetic needs thanks to the translocation of photosynthates from their symbiotic algae and complete their nutritional needs in terms of proteins and lipids capturing zooplankton. The relative importance of the two sources vary according to the individual species: within the corals of the genus Porites, for instance, P. astreoides is known to primarily use zooplankton for nutrition, whereas P. porites is capable of being fully autotrophic, zooplankton ingestion representing an insignificant component of its energy intake (Morri and Bianchi 1995). There is no universally accepted term for organisms that functionally lie between autotrophs and heterotrophs. The word mixotrophs seems etymologically adequate but is commonly employed for certain microorganisms that have the property of assimilating organic compounds as carbon sources but not as energy sources. A term recently crept into the coral reef literature is holobiont , coined to indicate the intimate symbiosis between the coral and its zooxanthellae (Yap 2007); there are examples, however, in which this symbiosis is not obligate, and the same species can live with or without zooxanthellae (Schuhmacher and Zibrowius 1984). Zooxanthellate corals have often been called polytrophs , but the term is both vague and ambiguous. It refers to the capacity to feed on multiple types of food, or to obtain nutrients in more than one way, and most so-called filter feeders are actually polytrophs in that they can incorporate both particulate and dissolved organic matter. As zooxanthellate (zoocyanellate, zoochlorellate) animals combine autotrophy and heterotrophy, they will be named “combo ” organisms in this chapter.

Global disturbances such as seawater warming affect differently autotrophs, heterotrophs, and combos, which is expected to cause shift in the trophic organization of the foundation species assemblages of animal forests. Change may propagate through the food web of the associate mobile consumers, upsetting both the detritus pathway and the benthic-pelagic coupling.

The bleaching event of 1998 caused the mass mortality of combo species (zooxanthellate scleractinians, alcyonarians, hydrocorals, Tridacna) in many tropical areas. In the Maldives, the relative proportion of the three aforementioned trophic strategies (autotrophy, heterotrophy, and combo) was visually assessed in the field with the plan view technique in replicated spots of about 25 m2 at 5, 10, 20, 30, 40, and 50 m depth along the reef profile (Morri et al. 2015). Surveys were carried out twice before the mass bleaching (in 1997 and 1998) and twice after (in 2012 and 2013). Trophic organization was synthesized in an index ranging from −1 (100 % autotrophy) to +1 (100 % heterotrophy), with 0 indicating absolute dominance of combos or equal proportion of autotrophs and heterotrophs. Results showed heterotrophy increasing with depth in both periods (Fig. 7), as expected from the gradual dimming of ambient light. In 2012–2013, combos returned dominant down to about 20 m depth, as they were before the bleaching. However, in 2012–2013 the proportion of combos was reduced in favor of autotrophs at all depths, and especially at 30 and 40 m, with respect to 1997–1998. Thus, although a phase shift from coral-dominated to algal-dominated reefs has not been evidenced in terms of community composition, the analysis of the trophic organization suggests that the food web basis has changed significantly. Thus, ascertained structural and compositional resilience may not necessarily vouch for functional resilience of the animal forest ecosystem (Duarte et al. 2015).

Fig. 7
figure 7

Change in trophic organization with depth in Maldivian coral reefs in years before (1997 and 1998) and after (2012 and 2013) the mass mortality of 1998. Trophic organization is synthesized in an index ranging from −1 (100 % autotrophy) to +1 (100 % heterotrophy), with 0 indicating absolute dominance of combos or equal proportion of autotrophs and heterotrophs (see text). Horizontal bars and shadowed areas represent standard errors around mean values (Data from Morri et al. 2015)

Full size image

9 Structural Complexity

Like their vegetal counterparts on lands, animal forests evolve following a successional path from early assemblages of pioneer species that settle directly on the bare rock to a dense canopy and high biotic coverage of the substratum. This process implies increasing structural complexity both in terms of species diversity and spatial entanglement. Major disturbances inducing mass mortality of foundation species inevitably lead to a decrease in the structural complexity of animal forests. Graham et al. (2015) calculated that structural complexity was one of the main factors favoring Indo-Pacific coral reef recovery after a major climate-induced bleaching event.

Structural complexity is the result of a long successional history of growth and stratification. The most mature animal forests consist of taller and more branched corals, more intricate and bigger sponges, etc., which can alter major current flows and particle retention. In contrast, immature animal forests have a smaller surface exposed to the major currents, and therefore their capacity for capturing carbon, nitrogen, phosphorus, and other elements is much lower: in synthesis, simplified animal forests can process less energy (Rossi 2013).

In coral reefs, three different stages of structural complexity have been recognized: (1) young or immature stage, (2) mature stage, and (3) regressive stage. In the young stage , reef communities are characterized by the dominance of reef builders (scleractinian corals and coralline algae); reef accretion and consolidation are encouraged and erosion is reduced. The mature stage represents a balance between coral growth and sediment deposition, with a comparatively lower abundance of scleractinian corals. Finally, the regressive stage has sparse living hard coral cover and high amounts of rubble and sand. Rowlands et al. (2016) used similar terms (juvenile, mature, and senile) to describe aggradation capacity, stating that high reef resilience is more common in mature reefs. Before the mass mortality of 1998, the coral reefs of the Maldives were mostly in young to mature stages. In 2006–2007, the state of Maldivian reefs was evaluated paying attention to the three ecological stages, assessing by the plan view technique the relative importance of three biogeomorphological descriptors : (1) living hard corals, (2) rubble and sand, and (3) coralline algae (Lasagna et al. 2010b). Surveys were done in the reef flat and in the slope in 16 replicated sites with different wave exposure.

A great number of the reefs studied in 2006–2007 were still in an ecological regressive stage, suggesting no recovery. Rubble and sand were widespread in all sites, whereas coralline algae, which contribute to their cementation, were comparatively infrequent. However, some of the reefs were in the young stage, and few in the mature one, thus suggesting some recovery (Fig. 8). The ecological stage depended on depth and exposure. The reef flat appeared in young stage in sheltered sites and in regressive stage in exposed to very exposed sites. Thus, wave energy in the flat acts essentially as a deterrent for coral growth. In the slope, the ecological stage appeared inversely related to wave energy: “young” in very exposed sites, “mature” in exposed sites, and “regressive” in sheltered sites. On the slope, therefore, wave energy was favoring reef recovery. The environmental regime is thus likely to play a major role in the resilience of animal forests, as far as the structural complexity is considered.

Fig. 8
figure 8

A summarizing scheme of the ecological stages (young, mature, regressive) observed in the coral reefs of the Maldives in 2006–2007, according to depth zone (reef flat and reef slope) and exposure (see text) (Redrawn and modified from Lasagna et al. 2010b)

Full size image

10 Constructional Capacity

Animal forests are shaped by autogenic ecosystem engineers , i.e., organisms that modify the physical environment with their own mass. Arborescent anthozoans and sponges may build a canopy more than 1 m high, thus providing new habitat for other species, either epibionts or mobile associates. Such a new habitat is defined “biologically mediated habitat ”, and the phenomenon is called “biological habitat provision ” (Cocito et al. 2000). When the autogenic ecosystem engineers possess calcareous skeletons, the deposit of a carbonate structure persisting after their death is called “bioconstruction ”, the term indicating either the process or the product (Bianchi 2001).

The best known bioconstructors are scleractinian corals, which are the main builders of both shallow-water tropical coral reefs and deep- or cold-water coral reefs. However, many other bioconstructors exist in the sea. Mediterranean coralligenous reefs, for example, are mostly built by calcified rhodophytes. Bryozoans, barnacles, serpulids, vermetid gastropods, oysters, mussels, and other invertebrates with calcareous skeletons and tests normally act as binders or secondary constructors but may even play the role of primary builders under certain circumstances (Bianchi 2001). Both coral reefs – at any depth – and coralligenous reefs result from the dynamic equilibrium between bioconstruction and destruction processes (by borers and physical abrasion), which create morphologically complex substrates where highly diverse benthic assemblages develop.

Mass mortality of bioconstructors following major disturbances may stop or hamper the process of bioconstruction, facilitate erosion, and leave place to soft-bodied organisms that frequently exhibit fast growth rates. Should this happen, the recovery of bioconstructors would be further delayed. The constructional capacity in Maldivian coral reefs following the mass bleaching and mortality of 1998 was evaluated in 2012 and 2013 at various depths down the reef profile and compared with the situation before the bleaching (1997 and 1998). Three categories were considered: hard-bodied organisms (mostly scleractinian corals, plus occasional large clams), which include all the primary and secondary bioconstructors; soft-bodied organisms (mostly soft corals, sponges and fleshy macroalgae), which have no construction potential per se but can act as bafflers , helping retain sediments within the reef; and non-biotic attributes (rock, sand, rubble), which evidently do not contribute to the bioconstruction. The substrate occupancy (in percentage) of these categories was visually assessed in the field with the plan view technique in replicate spots at 5, 10, 20, 30, 40, and 50 m depth (Morri et al. 2015).

The dominance of hard-bodied organisms at 5 m and 10 m depth appeared restored in 2012–2013, but the absolute values were lower than before the bleaching at 5 m (Fig. 9). In 2012–2013, the abundance of non-biotic attributes at 5 m was greater than in 1997–1998, whereas that of soft-bodied organisms at 10 m was lower. Hard-bodied organisms, soft-bodied organisms, and non-biotic attributes were similarly represented at 20 m and 30 m depth in both 1997–1998 and 2012–2013. On the contrary, at 40 m and 50 m non-biotic attributes were more abundant in 2012–2013 than in 1997–1998, to the detriment of soft-bodied organisms. This may be interpreted as the result of increased sediment (rubble and sand) production after the mass coral mortality. Mostly generated in the flat and upper slope, this coral-derived sediment eventually deposited on the terraces that characterize Maldivian reefs at those depths (Bianchi et al. 1997).

Fig. 9
figure 9

Change in constructional capacity of Maldivian reefs at different depths in years before (1997 and 1998) and after (2012 and 2013) the mass mortality of 1998. Constructional capacity is expressed by the % proportion (mean ± standard error) of hard-bodied organisms, soft-bodied organisms, and non-biotic attributes (Data from Morri et al. 2015)

Full size image

Significant constructional capacity resulted therefore limited to the shallowest depths (<20 m). Change in cover of different bioconstructor guilds was therefore evaluated between 1997 and 2015 (but in 2011) at shallow depths (5–15 m) in both ocean reefs (ocean-facing sides of atoll rims) and lagoon reefs (lagoon patch reefs or lagoon-facing sides of atoll rims). In particular, four bioconstructional guilds and one group of non-biotic attributes were considered: (i) primary builders are those organisms that build the reef framework and therefore assure reef aggradation ; (ii) secondary builders provide calcareous material to fill in the frame; (iii) binders are encrusters that consolidate the reef edifice; (iv) bafflers are soft-bodied algae and colonial invertebrates that, although not actively participating in the bioconstruction, help retaining sediment; and (v) non-biotic attributes (rock, rubble, sand) do not give any contribution to the bioconstruction. A bioconstruction potential index (BCP) was then devised using the following formula:

$$ \mathrm{B}\mathrm{C}\mathrm{P}={\displaystyle \sum_{\mathrm{i}=1}^{\mathrm{n}}\left({s}_{\mathrm{i}}{\mathrm{C}}_{\mathrm{i}}\%\right)\times {100}^{-1}} $$

where, n is the number of guilds (5, in this case), s i is an importance score assigned to the ith guild, and Ci% is the cover of the ith guild. The value of s i has been established at 3 for the primary builders, 2 for the secondary builders, 1 for the binders, 0 for the bafflers, and -1 for the non-biotic attributes. Therefore, BCP ranges theoretically from 3, in the unrealistic case of 100 % cover by primary constructors, to -1, when only non-biotic attributes are present and no bioconstruction is possible, the reef thus being prone to erosion and drowning.

Applied to the Maldives data (Fig. 10), BCP resulted highly correlated (R2 = 0.979) to total hard coral cover (which is evidently desirable) but provided clear threshold values to evaluate constructional capacity. Negative values imply no bioconstruction, and they characterized Maldivian reefs between 1999 and 2003 (lagoon reefs) or 2007 (ocean reefs). Values between 0 and 1 depict reefs virtually deprived of primary builders and therefore capable of constratal growth only. Reefs might not be able to keep up with the ongoing sea level rise. These values characterized Maldivian reefs between 2004 and 2010 (lagoon reefs) or 2008 and 2013 (ocean reefs). Values of BCP greater than 1 (corresponding to a hard coral cover >50 % in the Maldives) are indicative of superstratal growth , due to the relative abundance of primary builders, and characterized Maldivian reefs in 1997–1998 and again after 2012 (lagoon reefs) or 2014 (ocean reefs).

Fig. 10
figure 10

Time trend of BCP, the bioconstruction potential index, on shallow (5–15 m) ocean and lagoon reefs in the Maldives, 1997–2015. Negative values of BCP imply no bioconstruction, values between 0 and 1 depict reefs capable of constratal growth only, values greater than 1 are indicative of superstratal growth (see text)

Full size image

The coral reefs of the Maldives have therefore recovered most of their constructional capacity, even if hard coral cover on ocean reefs is still lower than pre-bleaching values. The other side of bioconstruction is obviously bioerosion (Hutchings 1986; Glynn 1997). The incidence of bioerosion in the destruction of Maldivian corals after the mass mortality of 1998 has not been studied; however, during our field work we had the opportunity to notice the abundance of clionaid sponges in dead massive corals, which might have contributed to the huge production of coral rubble recorded in 1999 (Lasagna et al. 2006).

11 Seascape

Forests on land are major landscaping elements, and the structure and dynamics of the mosaic of different vegetation (and hence habitat) patches are the core of landscape ecology . Animal forests in the sea are miniaturized mimics of vegetal forests on land but may nevertheless be considered as a prime example of submerged seascape . However, differences in biodiversity patterns between land and sea (Boudouresque et al. 2014) and, especially, lack of perception by common people of what thrives under the sea surface (Rovere et al. 2011) hampered the development of the discipline of seascape ecology until recently (Bianchi et al. 2005). Marine scientists are now aware that the application of landscape ecology concepts and techniques to the sea plays a central role in the study and monitoring of the multiscale processes that drive environmental change (Pittman et al. 2011). Terms such as benthoscape (relating to the seafloor) and reefscape (relating to coral reefs) were born, whereas no special terminology has been developed for the pelagic environment. Critical ecological thresholds are known to exist in the structural patterning of biogenic seascapes; exceeding these thresholds triggers abrupt phase shifts (Boström et al. 2011).

Seascape approaches integrate various levels of information from species identification to community structure and habitat characterization. They describe relationships between ecological processes and the spatial configuration of ecological mosaics. In the ecological literature, animal forests have been in turn defined as eco-ethological crossroads, biocoenoses, polybiocoenotic entities, assemblages, communities, community puzzles, and seascapes (Gatti et al. 2012; and references therein). Since animal forests are characterized by high heterogeneity, extreme patchiness, and coexistence of several ecological groups, a seascape approach seems to be the most reasonable solution for their characterization. Yet, few studies have applied landscape metrics to quantify coral reef resilience (Martínez-Rendis et al. 2016).

An important spatial pattern of animal forests is the increase of habitat three-dimensionality thanks to their stratified structure, which implies significant vertical elevation from the level bottom. At least three main distinct layers can be recognized: (1) upper layer , characterized by organisms with considerable (>10 cm) vertical growth, which forms the canopy; (2) intermediate layer , constituted by organisms with moderate (1–10 cm) vertical growth, which forms the understory of the forest; (3) basal layer , constituted by encrusting organisms, which protects and consolidates the substrate (Gatti et al. 2015b). Further layers can be recognized both at the highest and the lowest vertical extremes of the forest, namely, in the epibionts and the mobile biota associated to the canopy and in the borers and coenobites living in the interstices of the substrate, respectively.

Both natural and anthropogenic disturbances typically depress the 3D structure of marine forests (either algal or animal), thus causing seascape getting flatter (Bianchi et al. 2012). The destruction of the animal forests, in particular, would lead to an oversimplification of benthic ecosystems (Rossi 2013). The mass bleaching and coral mortality of 1998 did not change at first the reefscape of Maldivian coral reefs (Fig. 11). Notwithstanding the severe loss in living coral cover, the three-dimensional structure of the reef was preserved, dead coral colonies still being in place (Morri et al. 2015). By 2000, however, these dead coral colonies started to be destroyed by physical and biological erosion; by 2002, the three-dimensional structure of the reef was largely lost due to the breakage of dead colonies, which were reduced to rubble and sand (Lasagna et al. 2006). These loose detrital elements eventually deposited on reef slopes, smoothing the substrate and reducing reef rugosity. By 2004, a canopy of tabular Acropora was formed again, and the upper layer of the 3D structure was therefore reestablished (Lasagna et al. 2010a). However, both the intermediate and the basal layers remained severely underdeveloped (Fig. 2). In 2014, Maldivian reefscape looked yet very different (Morri et al. 2015) from what described before the bleaching (Morri et al. 1995; Bianchi et al 1997). Morri et al. (2010) argued that the reconstruction of the structural framework is a prerequisite for complete reef recovery.

Fig. 11
figure 11

Change in the Maldivian reefscape following the mass mortality of 1998. The dead coral colonies that remained initially in place, thus preserving temporarily the three-dimensional structure of the reef (1999), had been later reduced to rubble by physical and biological erosion (2002)

Full size image

12 Final Remarks and Recommendations

This chapter used a data series on Maldivian coral reefs as a model study case of pattern and timing of animal forest recovery after a major disturbance: the bleaching of 1998 that killed nearly 95 % of the foundation species of this particular animal forest (Bianchi et al. 2003). In this era of global change, coral reef resilience has been the object of many studies in different regions of the world ocean (Hughes et al. 2003, 2010; and references therein), but a major difficulty is finding the adequate empirical descriptors of resilience (Nyström et al. 2008). The Maldivian data set (Morri et al. 2015) is unique in that it considers an array of ten different descriptors , ranging from demographic parameters to ecosystem and seascape aspects. This provides the opportunity to compare different assessments of recovery and resilience (Table 1).

Table 1 Recovery and resilience of Maldivian coral reefs after the mass mortality of 1998, as estimated by yearly monitoring (till 2014) and ten different descriptors
Full size table

Sustained monitoring showed that Maldivian coral reefs reached complete or almost complete recovery when considering demography (recruitment, colony density and size, colony health) or synthetic descriptors like hard coral cover and diversity (in terms of species richness). Diversity, however, provided ambiguous results, as the total number of species was reestablished, but some previously common species has never been found again; recovery might therefore be considered unachieved. Diversity provided equivocal results also in one of the few studies that evaluated recovery of seamounts (Williams et al. 2010).

When considering descriptors at ecosystem level, recovery of Maldivian coral reefs was partial (community composition, seascape) or incomplete (trophic organization, structural complexity); only the constructional capacity appeared nearly reestablished. Consistently, resilience varied from 4 to more than 16 years (always remembering that recovery was rarely complete) according to the specific descriptor considered. Thus, although the risk of a structural phase shift seems avoided, Maldivian coral reefs are possibly suffering hysteresis. Marine animal populations and ecosystems are known to show some capacity of recovery but rarely to former levels of abundance (Lotze et al. 2011). Maldivian data series results are consistent with earlier studies estimating that the resilience of coral reefs after large mortalities is comprised between 5–6 years and more than 100 years (Bianchi et al. 2006). In the ample literature on coral reef recovery after disturbance, it has generally been assumed that recovery will eventually occur (Nyström and Folke 2001). On the other hand, it has also been observed that there may be critical thresholds beyond which resilience is lost (Mumby et al. 2007).

It is difficult to say to what extent the Maldivian lesson could apply to other animal forests. Resilience of ecosystems dominated by corals is low compared to most other marine systems (Williams et al. 2010). This chapter, however, showed that large differences in timing could, at least in part, depend on how resilience is actually measured. Notwithstanding the various attempts of rigorous definitions that followed one another in the last 40 years, the concept of resilience remains vague, varied, and difficult to quantify (Standish et al. 2014); similarly, there is no standard definition of recovery (Lotze et al. 2011). Resilience has even been defined as a nonconcept (Bianchi 1997), as there is no going back for species and ecosystems (Bianchi et al. 2012). After disturbance, animal forests are more prone to the invasion of alien species , which may replace the original components and further reduce the scope for returning to the original state (Gatti et al. 2015a). The highly species-rich animal forests in tropical shallow waters are supposedly less vulnerable to invasions, but this is not true. For instance, the Atlantic octocoral Carijoa riisei has invaded the Indo-Pacific (Calcinai et al. 2004), while the Indo-Pacific scleractinian coral Tubastraea micranthus has invaded the Atlantic (Sammarco et al. 2010). Little is known, for this respect, about deep- and cold-water animal forests.

Despite these conceptual difficulties, trying to estimate animal forest resilience remains of paramount importance for both science and management. The main take-home message from the Maldivian example analyzed in this chapter is the need of adopting several descriptors at various levels of integration. While tropical coral reefs are usually polytypic (Bianchi 2001), the animal forest canopies in temperate or cold waters are mono- or oligotypic , being shaped by just one or few species. This justifies that recovery and resilience are often estimated using only demographic descriptors (Teixidó et al. 2011; Santangelo et al. 2015). Population recovery is not necessarily indicative of ecosystem recovery (Lotze et al. 2011), as shown by the tabular Acropora example in the Maldives. Concentrating on the population structure and dynamics of a single or few dominant species may not be sufficient even in nontropical temperate forests, as change in community composition may persist after the apparent recovery of the canopy (Gatti et al. 2015a). Combining demography with several descriptors at the ecosystem level may be of help to get a reliable measure of animal forest resilience.

13 Cross-References