Experiments were carried out in June 2009 and January-February 2010 on a patch reef 2.5 km south of Carrie Bow Cay, Belize (16˚48.173'N, 88˚4.928'W) at 8.5 m deep (Figure 1). The site was well suited for the study because of high sponge abundance, a diverse octocoral population, and signs of common sponge-octocoral inte- ractions.

The branching sponges and octocorals used in the experiments are representative of species common to many Caribbean reefs. All sponges belong to the class Demospongiae: Desmapsamma anchorata (Poecilosclerida, Desmacididae) and Iotrochota birotulata (Poecilosclerida, Iotrochotidae) commonly overgrow octocorals; Niphates erecta (Duchassaing & Michelotti) (Haplosclerida, Niphatidae) and Ptilocaulis walpersi (Duchassaing & Michelotti) (Halichondrida, Axinellidae) are rarely found epizoic on octocorals. N. erecta has been observed grow- ing on the sea fan Gorgonia ventalina (Linnaeus) [5] but never on branching octocoral species.

As morphology influences the nature of species interactions [26] , our four selected species span a range of shapes and distributions: Desmapsamma anchorata (Figure 2(B) and Figure 2(C)) is encrusting to arborescent, widespread, found from near-shore to oceanic habitats and from calm deep water to exposed shallow reef habi- tats [21] [39] [40] . D. anchorata grows fast, is easily torn, exudes slippery mucus upon tearing [40] and produces a variety of secondary metabolites and anti-predatory toxins [41] [42] . Its growth is described as weedy, growing substantially faster than other branching sponges [1] [43] . Iotrochota birotulata is ramose, has a tough, com- pressible consistency and a spiky surface; it can be covered with symbiotic zooanthids [44] [45] (Figure 2(A) and Figure 2(D)), which are reported to repel predators [46] and increase the sponge’s chance of survival [47] .

Niphates erecta and Ptilocaulis walpersi too are branching sponges that are widely distributed in the Caribbean and found on reefs and other hard bottoms [48] [49] . Their growth forms and colors are variable. N. erecta can be erect ramose, massive, or encrusting. Its surface is conulose, its consistency, compressible and tough and is frequently found covered by the zoanthid Parazoanthus parasiticus [40] [50] [51] . N. erecta differs from P. walpersi in that it is considered chemically undefended or poorly defended, although it has been reported to encroach on and overgrow the coral Madracis mirabilis (Lyman) [34] ; in contrast to other chemically undefended or poorly defended sponges, N. erecta grows slowly [11] . P. walpersi is chemically defended, but deters generalist fishes to a lesser degree than other chemically defended sponges [11] . P. walpersi morphology is characterized by an erect, finger-shaped body with dichotomous branches; it can grow also encrusting or fan-shaped, its surface ornamented by spiny, or round and blunt, or spatula-shaped projections. Although not frequently found associated with octocorals, P. walpersi is found to overgrow stony corals and is a considered a strong space com- petitor [52] .

The octocorals (Gorgonacea) chosen for the experiments have different, distinct morphologies. Most branching species in the Caribbean have a semi-rigid scleroproteinaceous axis [53] that, if exposed after tissue death, becomes a substrate over which sponges and other organisms may grow. Antillogorgia americana (Gorgoniidae) colonies are tall with long regular plume-like branches; they are slimy to the touch and can produce copious amounts of mucus containing predator-deterrent compounds [54] . Eunicea flexuosa (Plexauridae) colonies range from candelabrum-like to bushy, with a surface that is smooth or bumpy with many calices [54] . Plexaurella nutans (Duchassaing & Michelotti) (Plexauridae) similarly has candelabrum-like colonies, with a branch surface that has the texture of fine sandpaper; it has not been observed with sponge overgrowth [54] .

Results are for 12 replicates per treatment, testing three different physical properties of the sponge (live, dead, tissue mimic) on the octocoral Antillogorgia americana. Initial observations, on the first 5 days, were consistent when testing the physical properties of the sponges, for all samples when different sponge fragments were in contact with the octocoral tissue responses observed ranged from discoloration to tissue death. Generally, the time of exposure, 1 - 8 days, significantly influenced the response on the octocoral (F_2,136 = 22.82, P < 0.0001), as well as the physical property of the sponge fragment in contact with the octocoral substrate (F_2,136 = 4.54, P = 0.012) (Figure 4).

On the attachment response, Desmapsamma anchorata and Iotrochota birotulata differed for both the live and the dead fragments (t₁₃₆ = −2.15, SE = 0.85, P > 0.0329). The range in responses was different for the live versus inert sponge fragment properties: the responses to live tissue differed from tissue mimics (t₁₃₆ = 2.96, SE = 0.83, P = 0.0036), and there were no-significant differences between the tissue mimics and dead sponge fragments (t₁₃₆ = 0.82, SE = 0.78, P = 0.415) (Figure 4).

The attachment response of the live sponge fragments varied with time of exposure and the interacting species. After eight days, 25% of D. anchorata fragments were attached to A. americana. In contrast, 92% of the I. birotulata fragments in contact with the same octocoral species caused tissue death but did not become attached. Tissue death occurred faster when the tissue mimic was applied than in the cases of live-sample pairings; we also noted that more than 50% of the dead fragments were lost before the end of the experiment (Figure 4).

By species-specific effects overgrowth-deterrence is meant that the attributes of a given species, either the octocoral or the sponge involved act so as to deter the sponge attachment on the octocoral substrate. When different live sponge-octocoral species pairings were tested, the results mirrored the natural (non-experimental) situation where common epizoic species attached more often than rarely epizoic species (F_2,182 = 8.16, P < 0.0001). The response was significantly influenced by the interaction effect of sponge species and the octocoral species tested (F_2,182 = 6.9, P = 0.0002), and the time of exposure for the paired organisms (F_1,182 = 79.88, P < 0.0001).

For the two sponge attributes tested, we found that sponge-octocoral interaction within one category (common or rare epizoic) had similar responses, such as the effects of Desmapsamma anchorata and Iotrochota birotulata on the octocoral Antillogorgia americana (t₁₈₂ = −0.26, P = 0.0766), however the response effects across different categories differed (Figure 5). Comparing the effects of attachment and non-attachment of commonly epizoic D. anchorata and I. birotulata on the ocotocoral A. americana, compared to rarely epizoic Niphates erecta and Ptilocaulis walpersi on the same octocoral were significantly different (t₁₈₂ = 2.53, P = 0.0123 and t₁₈₂ = 2.85, P = 0.0049). Although discoloration and tissue death effects are present in both categories (Figure 5 and Figure 6), the ultimate response of attachment, leading to sponge overgrowth, was completely absent in the sponge- octocoral pairings that do not naturally occur (Figure 6).

In contrast to the previous experiment, where the species pairings were tested for species-specific effects on live substrate only, this experiment measures the ability of a sponge to attach on varying octocoral substrate, testing for increased susceptibility to sponge colonization in the absence of live-tissue on the octocorals more. The re- sponse observed on live octocoral tissue, characterized by discoloration and death of the tissue and followed by attachment was influenced by the interaction between sponge and octocoral (F_1,211 = 23.4, P < 0.0001) and the number of days of the interaction (F_1,211 = 51.76, P < 0.0001). In the case of sponge-octocoral associations be- tween species that are not commonly epizoic, the attachment success was significantly different for live tissue versus bare octocoral axis (t₂₁₁ = 4.84, SE = 0.53, P < 0.0001). The response was different for Desmapsamma anchorata tested on live Plexaurella nutans where attachment was not observed; however, the attachment response was greater on to the bare axis of P. nutans (Figure 7), than in the D. anchorata and Antillogorgia americana species pairing (Figure 5), indicating that the bare axis of P. nutans can be more susceptible to overgrowth than the bare axis of A. americana. Within 3 days of the start of the experiment, D. anchorata attached to A. americana, completing a full circle around the octocoral axis (Figure 2(C)). This result was different from the initial tests, were even after eight days of the experiment there was no attachment or overgrowth of D. anchorata on the live P. nutans, although signs of discoloration occurred (Figure 7). Results on the octocoral reaction comparing Desmapsamma anchorata and Iotrochota birotulata on the live and bare tissue of Antillogorgia americana and Eunicea flexuosa indicate that attachment success is higher and faster on A. americana than on E. flexuosa, and that the D. anchorata is more successful attaching on the bare substrate, than on the live tissue (Figure 5).

Preliminary measurements on growth indicate that different octocoral species, in the sponge-octocoral pairings monitored, appeared to influence the growth of the sponge for Iotrochota birotulata, but not for Desmapsamma anchorata. The mean growth of I. birotulata was different on Antillogorgia americana (M = 1.57, SE = 0.68), when compared to its growth on Eunicea flexuosa (M = 0.68, SE = 1.05), t₂₄ = 2.55, P < 0.05 (Figure 8). How-

ever, for the faster growing D. anchorata on A. americana, there was no significant difference (M = 2.32, SE = 2.44), compared to E. flexuosa (M = 3.31, SE = 3.59), t₂₅ = 0.85, P > 0.05 (Figure 8).

Differential growth rates measured for the two sponge species resulted in contrasting overgrowth differences. After six months, initial fragments of D. anchorata measuring an average initial area of 7 cm² grew from 7.8 cm² to 83 cm²; and fragments of I. birotulata measuring an average initial area of 9 cm² grew more moderately, from 10 cm² to 23 cm², varying from their initial length along the axis from 5 cm to 3 - 14 cm (Figure 8).

The physical response observed on the octocoral tissue in the sponge-octocoral interactions varied across the different properties of the sponges that were in direct contact with its surface. In general, the effect of discoloration that leads to tissue death is observed across most interactions, with attachment only possible for live sponge fragments of species that naturally overgrow the octocoral. Considering that the sequential response is similar across all interacting sponge type tested helps understand how a neighboring organisms, being inert or alive, can negatively affect the integrity of the octocoral and when attachment occurs both their living tissue and the supporting axis becomes a substrate. In our experiment, although the number of days differed slightly (Figure 4), the sponge mimic had as great an effect in causing tissue death as the living sponge tissue, when in contact to the octocoral, indicating that, in part, the effect was primarily physical. In situ, 50% of the dead sponges were lost before the end of the experiment; dead sponge fragments of Desmapsamma anchorata tissue tended to disintegrate and fragments of Iotrochota birotulata probably fell because these fragments were more rigid and could not resist the drag from the water column; the commercial sponge skeleton was of a soft, elastic consistency and remained in the pairing longer, tending to disintegrate eventually while still negatively affecting the octocoral substrate (Figure 4).

The physical effects of direct contact can cause abrasion and damage on the octocorals base, effects that are conducive to high octocoral mortality [10] [63] . In general, sponges benefit from these interactions by gaining structural support and protection from predators [16] [62] [64] , while octocoral axes can become debilitated [5] [10] and susceptible to other foulers. Similarly, this effect has also been observed on scleractinian corals, their susceptibility to being overgrown increases when their polyp-covered surface is affected by smothering action [65] .

The ability of a sponge to attach to an octocoral is influenced by species-specific effects, this influencing the fre- quency of sponge-octocoral interactions observed on a reef. The factors considered in our experiments, sponge species, octocoral species, common versus rare epizoics, as well as the number of days of the interaction, determined the observed outcome. Of particular interest in this part of the study was the response effect of sponge- octocoral species pairings that are rarely epizoic or overgrown on a reef, such as test trials of Eunicea mammosa (Lamouroux) in contact with Iotrochota birotulata, where the octocoral tissue elicits a response that resembles an inflammation on the area of contact. A different reaction was observed on the rarely overgrown Plexaurella nutans when in contact with Desmapsamma anchorata; here the octocoral tissue changes color, from yellow to black (Figure 2(B)). In both of these cases there was no sign of sponge attachment, even after 10 days of inte- raction (Figure 6 and Figure 7).

Common epizoic sponges like Desmapsamma anchorata and Iotrochota birotulata are rarely found to overgrow certain branching octocorals. The question on how the octocorals’ susceptibility to overgrowth varies in function of loss of the integrity of its substrate was tested. Different from the octocorals deterrence observed when the sponge was in direct contact to live octocoral tissue, we observed that the response changed resulting in sponge overgrowth when the scleroproteinaceus axis of Plexaurella nutans had been compromised (Figure 7). For sponge species that overgrow, attachment of a fragment is always faster when in contact with bare octocoral axis rather than the polyp-covered surface. Therefore, an increase in the susceptibility of the octocoral substrate allows for an epizoic sponge, like D. anchorata, to attach and grow not only on octocoral species on which it naturally occurs, but also on others, although the process is slower.

In a dynamic reef context, sponge-octocoral interactions leading to attachment and overgrowth are facilitated by a compromised host; smothered or dead octocoral tissue being more susceptible to overgrowth than a healthy one, and then by wave action or a limited time of interaction. Sponges growing on octocorals are known to have an effect on their hosts [62] that is likened to an immunological response [2] [16] [66] that varies from species to species. Visually, this response is obvious by the exudates of slime, common for both tested branching octocorals, Antillogorgia americana and Eunicea flexuosa, when exposed to the sponges Desmapsamma anchorata or Io- trochota birotulata. One exception is the nature of the response seen in Carijoa riisei (Duchassaing & Michelotti) when overgrown by D. anchorata; here their interaction leads to formation of a continuous collagen sheet on the octocoral surface, with no exudate [2] [16] .

The recorded sponge growth from our experiments simulates natural conditions where sponges growing over octocorals on the reef are exposed to breakage and predation [20] [56] [67] . As a result, their growth rate can be negative or positive. Growth of sponges is species-specific [19] and affected by the nature of the substrate over which they grow. Comparing growth of one sponge species tested on 2 different octocoral substrates: Differences in mean growth were observed in Iotrochota birotulata when growing on Antillogorgia americana versus on Eunicea flexuosa, but not in Desmapsamma anchorata, which can grow four times faster than I. birotulata; differences in its growth rate on the two tested octocoral substrates were not significant. Desmapsamma anchorata’s attribute of spurred growth also means that it is frail and breakable, increased wave action can break it off its overgrowing substrate [3] [60] , in contrast to Iotrochota birotulata that grows slower but is more persistent.

Next to interacting species influencing sponge growth, environmental aspects such as the position of the animal in the water column [8] [57] the speed of water flow [58] , food content, resource allocation [68] [69] , as well as associated fauna or symbionts [62] [70] [71] are important in this process; furthermore sponge growth is also influenced by the ecological effects of predation [56] .

Differences across other taxa that interact with sponges are also noted, not all of them negative. For instance in sponge-coral associations, sponges have been observed to enhance the survival of the corals [38] or to prevent other organisms from fouling them [72] . However, in the case of corals, overgrowth susceptibility increases when the coral surface is affected by smothering [65] .

In sponge-octocoral interactions, sponges tend to impact branching octocorals negatively by their growth; this effect is best described as ‘structural parasitism’ or lianas [73] where sponges topple the octocorals over which they grow [8] . Going back to Stebbing’s definition of overgrowth, it is important to take note of organisms that extend their edge over others, as once this step is taken, the aggressive organism has the potential to overgrow the other [36] for instance, a sponge like Desmapsamma anchorata, known to have substantial growth rates, can double its length along an octocoral axis in only 6 months.