A Case Study in Maritime Heritage Ecology: Understanding How Structural Changes to the 1898 Shipwreck Portland Affect Biological Diversity and Colonization

Shipwrecks are irreplaceable cultural and historical resources, and they also serve as biological habitats. The physical structure of a shipwreck provides habitat for hard-bottom organisms, including sessile invertebrates (anemones, sponges) and protected areas for fish. The structure of a shipwreck is influenced by cultural and natural site formation processes over time. A key unanswered question is: How do site formation processes and changes in biological communities (ecological succession) influence one another? We studied the shipwreck Portland in Stellwagen Bank National Marine Sanctuary to answer this question, working within the interdisciplinary framework of Maritime Heritage Ecology. Analysis of sonar and video data from 2002 to 2010 and 2019–2021 revealed substantial structural changes to the bow, freight deck, and stern. Site formation on the bow and stern was clearly anthropogenic, as shown by entangled fishing ghost gear. The assemblage of sessile invertebrates in each of the three shipwreck areas also saw changes in community structure, including changes in the relative abundance of species. We also observed changes in the fish community, but these are more likely the result of regional climatic change. Overall, our study showed that ghost gear is a significant threat to Portland and other shipwrecks in Stellwagen, and that site formation drives changes in the biological community over time.


Introduction
Shipwrecks are multi-faceted, dynamic databases that may be studied as artifacts themselves, as carriers of artifacts, as microcosms of maritime cultures and associated systems, or combinations of all the above (Muckelroy 1978;Gould 1983;Murphy 1983;Oxley and Keith 2016). They are also valuable biological habitats and many function as island-like habitats. Shipwrecks are mishaps and are not supposed to exist. As mishaps, they have been removed from their systemic, use-context (transportation) to a non-use, archaeological context in a maritime environment without planning, preparation, or purposeful placement (as opposed to ships scuttled or sunk as artificial reefs) (Gibbs and Duncan 2016;Meyer-Kaiser and Mires 2022).
After deposition on the seafloor, shipwrecks increase habitat heterogeneity, cause longterm shifts in the benthic community composition, and lead to higher biodiversity both on their surfaces and in the surrounding sediments (Work et al. 2018;Bałazy et al. 2019;Casoli et al. 2020;Meyer-Kaiser and Mires 2022). In some cases, shipwrecks can expand the distribution of a species, for example by hosting a species at the edge of its geographic range or by serving as stepping-stones for invasive species (Paxton et al. 2019;Soares et al. 2020). For instance, sessile invertebrates and reef fish use shipwrecks as habitats. They rely on the structure of the wreck for shelter, elevation off the seafloor, and a solid surface to live on. Shipwrecks provide microhabitats that serve as shelter and nursery areas for fish. By settling on a shipwreck, sessile invertebrates enjoy elevation in the benthic boundary layer that facilitates suspension feeding. These species also attract predators, which may be pelagic or anthropogenic (Meyer-Kaiser and Mires 2022). This biological colonization is an example of a natural transformation (n-transform), which is influenced by other n-transforms and anthropogenic impacts, or cultural transformations (c-transforms), such as fishing or diving activities.
There is a large knowledge gap, however, in this understanding of the reciprocal influence of shipwreck structure and biological community composition over time. Yet, understanding these interactions is critical for management of biological and cultural resources. This study leverages an unprecedented two-decade dataset for an interdisciplinary investigation in Maritime Heritage Ecology (Meyer-Kaiser and Mires 2022).
Maritime Heritage Ecology (MHE) is an interdisciplinary research framework that aims to understand the interactive biological, natural, and anthropogenic factors that drive site formation processes and answer critical management questions for Underwater Cultural Heritage (UCH) (Meyer-Kaiser and Mires 2022). Using this approach, we asked the research question: How do natural and cultural site formation processes and biological community composition reciprocally influence one another over time on the archaeological remains of the 1898 shipwreck, Portland, in Stellwagen Bank National Marine Sanctuary (SBNMS)?
In order to address this question, we focused on three areas on the Portland shipwreck site with greatest structural changes-the port bow, the freight deck, and the fantail stern. We first provide a brief background on Portland's history and its search and discovery to provide context for our methodology, which compared data sets between 2002 and 2010 and 2019-2021 missions. We provide analysis of the structural and biological transformations between these data sets, specifically discussing how ongoing anthropogenic factors and subsequent c-and n-transforms have impacted biodiversity on Portland's structural features through time.

History of the Paddle Steamer (PS) Portland
Built in 1889, PS Portland was a 91 by 20 m (m) (291 feet [ft] by 65 ft) side-wheel paddle steamer powered by two Bath Iron Works boilers, a single cylinder, and 9.3 m (30 ft) high walking beam (Lawrence et al. 2015, 27-29). It was a night boat that ran between Boston, Massachusetts and Portland, Maine, and for only a dollar one-way (approximately $36 today). Travelers boarded the steamship and descended down ornate staircases into the 70 m (225 ft) Grand Saloon with glass-domed skylights and electric chandeliers. They could enjoy beverages or food on mahogany sofas or chairs in red plush velvet carefully arranged on embroidered carpet. They could find different forms of recreation or relaxation through the trip, and if they could afford it, at night they could sleep in one of the 514 white-pine berths in the 167 cherry-paneled staterooms (Lawrence et al. 2015).
On Saturday, 26 November 1898, Portland left Boston for Portland with possibly 200 passengers and crew onboard. No one knows the exact number, as the only manifest was aboard the ship. Despite a telegram carrying reports of a severe storm moving up the coast, the steamship left at 7:00 pm as scheduled. The storm, which would be thereafter known as the Portland Gale of 1898, soon caught the steamship. The final sighting of the Portland was at 11:00 pm by a passing vessel, which reported that the ship appeared to be struggling against the seas (Lawrence et al. 2015;Milmore 2019). There were no witnesses nor survivors to tell of the ship's sinking, but bodies and pieces of wreckage started to wash up on the shores of Cape Cod the next day and for weeks thereafter (Lawrence et al. 2015;Milmore 2019).

Search and Discovery of the Portland Shipwreck
The loss of Portland made headlines around the nation, and the search for the vessel and any survivors began immediately, despite the only passenger manifest being aboard the ship. This created confusion in accurately accounting for all passenger that still remains, but produced a lasting legacy that two manifests were kept for passenger ships thereafter: one for the steamer and one for the port of departure. Throughout the twentieth century, the loss and search for Portland was a popular subject for historians and shipwreck enthusiasts relative to other shipwrecks in the New England region. Authors compiling shipwreck stories of New England or the Northeast were sure to include, and often highlight, the story of Portland (Eames 1940;Snow 1943Snow , 1949Snow , 1960Quinn 1973;Fish 1989). Near the centennial of the steamship's sinking in the late 1990s, there was a renewed interest in Portland's story (Freitas and Ball 1995;Quinn 1996;Bachelder and Smith 1998;Bates 2000), but of course they could only speculate on the shipwreck's fate until its identity and location was confirmed in 2002.
It was in 1989 (a century after the ship's construction) when Portland was actually located by avocational shipwreck surveyors, John Fish and Arne Carr, and WHOI oceanographer, Richard Limeburner. Limeburner calculated hypothetical locations based on 1898 ocean currents and the time recorded on several pocket watches that washed ashore with some of the victims. Based on these calculations, Fish and Carr located a shipwreck in approximately 156 m (500ft) of water. Due to the quality of their data, however, confirmation of this wreck's identity had to wait until 2002 when they worked with SBNMS (as the site was now within the sanctuary's boundaries) to conclusively identify the shipwreck as Portland (Lawrence et al. 2015;Milmore 2019;Fish and Carr 2020;Mires et al. 2020).
The Portland shipwreck rests in Gloucester Basin within Stellwagen Bank National Marine Sanctuary. Established in 1992 at the mouth of Massachusetts Bay, SBNMS is the United States' 10th National Marine Sanctuary and protects more than 842-square miles of Gulf of Maine waters (Fig. 1). The sanctuary has an explicit mandate to protect all natural and cultural resources, including an estimated 200 historic shipwrecks: 47 of which  1 3 resources is complicated by the fact that SBNMS allows fishing activities as part of its original charter. The Gloucester Basin, where Portland sits, is a deep, muddy basin and prime area for fishing activities, such as gillnetting and bottom trawling. The water is cold (approximately 8 degrees Celsius) and generally quite turbid and strongly influenced by currents.

Materials and Methods
This study leverages two data sets from 2002 to 2010 and 2019-2021 missions to provide analysis and characterization of the changes and site formation on the Portland shipwreck site.

Baseline Characterization (2002-2010)
From 2002 to 2005, SBNMS archaeologists and staff conducted archaeological surveys on the Portland shipwreck site to provide a baseline characterization. They used a combination of side scan sonar, ROV video, and photomosaic imagery to provide assessments of the ship's condition (Fig. 2). They observed a diverse group of benthic and pelagic species in residence on and around the shipwreck: red and white anemones (Metridium senile and Urticina felina) and sponges (Mycale lingua, Polymastia robusta, and Polymastia mamillaris) encrusted the wreck's structure and Atlantic cod (Gadus morhua), pollock (Pollachius virens), and cusk (Bromse bromse) were observed throughout the ship's structure (SBNMS 2005). They also observed anthropogenic impacts, such as nets and lines floating, hanging, and entangling Portland's structure in many places. This ghost gear inhibited further documentation in many areas. Analysis of site formation processes was qualitative, with the focus on describing degradation of structural features and movements of artifacts on the main deck where researchers had access (SBNMS 2005).
From 2005 to 2010, they conducted near annual side scan operations to monitor the site, bringing a synthetic aperture sonar imagery for high-definition imagery in 2010. However, these images cannot provide information on biological diversity and changes in species. From 2010 to 2019, there were no archaeological or assessment surveys undertaken on the shipwreck site.

Assessment and Monitoring Surveys (2019-2021)
To rectify this situation, in 2019-2020, NOAA's National Ocean Service (NOS) and the Office of National Marine Sanctuaries (NMS) funded an interdisciplinary team from Woods Hole Oceanographic Institution (WHOI), Marine Imaging Technologies (MITech), and SBNMS to return to Portland and characterize both biological and anthropogenic changes and impacts to the ship's structure. Using a new cinema-class ROV with 40,000 Lumens of lights and up to eight cameras (including high-definition, large format and panoramic cameras) developed by Marine Imaging Technologies, we recorded imagery from 85 to 90% of Portland capturing standard-and high-definition imagery and video that was used for analysis and development of a high-definition 3D model of the shipwreck. In 2021, SBNMS partnered with Klein Associates and MIND Technology to acquire new high-definition sonar imagery with a prototype side-scan sonar that was tested on the shipwreck. All of this data was then processed for biological and archaeological analysis.

Analysis of Data Sets
For the biological analysis, identification and quantification of fish and invertebrate communities required specific methods to reduce bias and achieve accurate data given the specific and varied characteristics of pelagic versus benthic organisms. Video recordings were broken into segments when a storage drive became full (usually after 45-50 min of run time), and that segment became a discrete sample for analysis for the fish communities. The fish belonging to each species in and around the shipwreck and immediate area were counted. These numbers were used to calculate proportions of the community represented by each species in each video segment, with proportions averaged between video segments from a given year. This methodology reduces biases introduced by fish movement, such as double-counting (Ross et al. 2016). For invertebrates, however, video recordings were converted to frame grabs anytime the shipwreck surface came into clear view, whether or not there were any invertebrates in view. This approach reduced the risk of, and thus preventing data bias from, counting artificially high abundances of invertebrates.
Biological communities in each area (bow, freight deck, stern) were analyzed separately. Differences in community structure of invertebrates were quantified using multivariate PERMANOVA tests with unrestricted permutation of the raw data and Type II sum of squares in Primer v7 (Anderson et al. 2008). These differences were visualized using nonmetric multidimensional scaling in Primer.
For the archaeological analysis, 2019 and 2020 video footage were converted to still imagery by frame grabs. All 2019-2020 frame grabs and photographic documentation were compared to previous video and photographic evidence from 2002 to 2004. This provided an initial qualitative assessment of changes on Portland's structure, artifacts, and surrounding matrix. To establish scale and spatial context for metric analysis of the shipwreck, a mixed-methods approach of 3D modeling and side scan sonar imagery was used. Volumetric 3D models of portions of the shipwreck site were produced after all images were pre-processed to even out exposure, color and image quality and collated into numerous groups that could be built up as "mini models." Metashape software was used to align and merge point clouds of approximately a dozen mini models.
High-definition side-scan imagery-acquired by a Klein Associates prototype 4 K-SVY side scan sonar running at 600 kHz-was processed by Klein Associates and MIND Technology using Sonar Pro. Raw side scan files were converted into mosaics, and then range markers were added to provide a scale for measurements and aid in the creation of a site map. Next, we used the sonar images from 2003, 2010, and 2021 to quantify rates of deterioration and change. We encircled portions of the shipwreck that had deteriorated (e.g., areas where the upper decking was missing) and found the surface area of deteriorated areas using ImageJ. Scale bars in each sonar image provided a meter-to-pixel conversion.
Finally, we used ImageJ to determine the area of intact remaining structural remains (and therefore available habitat) in sonar images for an integrated analysis (combining analysis of cultural and natural transforms). Grayscale conversion allowed pixel-by-pixel gray values to serve as a proxy for sound reflectivity and therefore presence/absence of wood. Using the straight-line tool in ImageJ, we selected a transect from the funnel running forward to the bow in each sonar image (2003,2010,2021). The gray value of pixels in the forward hatch was used to determine a threshold value for empty space in each image. The hatch was an opening in the upper deck, so pixels that were lighter than the hatch were considered to represent wood (either decking or deck beams). The proportions of pixels above the threshold value were calculated for each year and used to show the percent of available habitat.

Site Formation Processes
Site formation processes are the causal mechanisms that move artifacts and structures from their systemic, or use context, to their archaeological context where further transformations can occur, and are generally divided into natural transformations (n-transforms) or cultural transformations (c-transforms). Originally put forth by Schiffer and Rathe (1973), the theoretical framework was applied to maritime archaeology when Muckelroy (1978) created a model that illustrated the various n-transforms and c-transforms on a shipwreck. Muckleroy's original model has evolved and been refined at different times since its original publication (Ward et al. 1998;Stewart 1999;Martin 2011;Gibbs and Duncan 2016). Studies have usually focused either on n-transforms, especially on geomorphic and sediment processes (Ford et al. 2016;Keith and Evans 2016;Quin et al. 2016) and biological processes (Gregory 2016;Perasso et al. 2022); or c-transforms such as fishing and diving (Brennan et al. 2016;Gibbs and Duncan 2016;Siciliano et al. 2016).

Maritime Heritage Ecology as an Interdisciplinary Framework
There is limited literature or understanding on how "site formation processes and ecological succession influence one another" (Meyer-Kaiser and Mires 2022). The interdisciplinary framework of MHE emerged from the 2019-2020 survey missions to address this knowledge gap. During the 2019-2020 survey, operations were conducted by a small, interdisciplinary crew, allowing for constant communication and cross-disciplinary information exchange between scientists in biology and archaeology. This research environment brought to light many challenges for integrated archaeological-biological collaborations from research design, theory, and methodologies to language and post-field work goals and publications (Meyer-Kaiser and Mires 2022). As Meyer-Kaiser and Mires (2022) note: Addressing these research challenges will require ecologists and archeologists to work in strong interdisciplinary teams. Frequent, genuine dialog and good-faith cooperation can mitigate cultural and training barriers. There is especially a ripe opportunity for investigations of mid-level theory…likewise, there is a real opportunity for researchers and managers to bring knowledge of ecological succession and site formation processes (both mid-level theories) into policy considerations.
Research studies range in theoretical frameworks from low-level (observations and descriptions); mid-level (attempts to account for patterns between variables and sites); and high-level (generalizations that can explain major phenomena). Each level provides a foundation for the next. MHE combines archaeological site formation and biological succession to develop mid-level theories to address (1) how do site formation processes and succession influence one another and (2) what is the net effect of UCH on biodiversity. UCH can have immediate and long-term impacts for marine ecology by providing habitats for hardbottom species (Paxton et al. 2019;Hamdan et al. 2021), altering community composition around the site (Work et al. 2008;Bałazy et al. 2019), and by facilitating the spread of invasive species (Soares et al. 2020). Additionally, the MHE framework emphasizes communication and policy, as understanding the reciprocal influence of site formation and ecological succession will be essential for managers and researchers of UCH. In this study, we used an interdisciplinary MHE approach to understand integrative biological and archaeological changes to the Portland shipwreck over time.

Results
Changes in Portland's structure and structural integrity were immediately evident in 2019 and were documented in greater detail in 2020 (Fig. 3). There was an increase in ghost gear, with a new net on the port bow and increase in deterioration of the freight deck. The greatest impact was to the stern, where the fantail timbers and railings were ripped off and resting on the seafloor with fishing line draped around the features (Fig. 3). Artifacts that were noted in the earlier projects had either moved or fallen into new locations or were not visible. Figure 3 illustrates anthropogenic impacts and c-transforms on the shipwreck through side scan imagery from 2002, 2010, and 2021.
The structural changes in the Portland site were accompanied by changes in the biological communities over time. The fish community was dominated by Atlantic cod, pollock, and Acadian redfish (Sebastes fasciatus). Cusk was found in small numbers on the Portland in most years, while other species were only found occasionally. The proportion of each species in the community fluctuated over time (Fig. 4). The starkest difference occurred in 2006, when the entire fish community was dominated by Atlantic cod. In subsequent years (2009-2020), cod represented much smaller proportions of the community, which was dominated by Acadian redfish.
For benthic invertebrate communities, we focused on results on port bow, freight deck, and stern areas of the shipwreck, which have significant c-transforms that impact the n-transforms, providing a model to show the interactions between site formation processes and biological community composition. Differences in the community structure of invertebrates were apparent for the freight deck, the port bow, and the fantail (PERMANOVA tests for freight deck (df = 6, pseudo-F = 3.76, p < 0.001), port bow (df = 7, pseudo-F = 3.40, p < 0.001), fantail (df = 1, pseudo-F = 5.28, p = 0.02); Fig. 5). In all cases, post hoc pairwise tests showed significant differences between sampling years prior to 2009 and 2019-2020.

Port Bow
One of the first and most obvious signs of anthropogenic impacts noted in 2019 was the appearance of a trawl net on the port side of Portland's bow. In 2004, SBNMS archaeologists reported that "Ghost gear drapes portions of the Portland's bow, starboard side" with lines cutting into the ship's stem post but did not mention a net or any ghost gear on the port side (SBNMS 2005). The 2010 side-scan image (Fig. 3) shows that the net had not been deposited on site, providing a terminus post quem (TPQ) of 2010 and a terminus ante quem (TAQ) of 2019, when first observed. The portside net is also highly visible in the 2021 side-scan image (Fig. 3). A rendered 3D model (Fig. 6) of the net provided basic measurements. Although the shape and sizes likely shift due to dynamic environmental conditions, like direction and speed of the tidal currents, on average, the net is 8 m high (25 ft) off the seafloor and 2.5 m (8 ft) at its widest point with a visible surface area of approximately 41.25 m 2 (132 ft 2 ).
Portland's port bow saw a steady decline in the anemone Metridium senile over time during the study period 2002-2020 (Fig. 7). This anemone covered the port bow almost completely in 2002 but was almost completely absent by 2020 (Fig. 8). A parallel decline in the sponge A. infundibuliformis was observed over the period 2004-2020, while the flat, encrusting species Halichondria panicea increased 2004-2020.

Freight Deck
Between 2002 and 2010, Portland's freight deck had limited structural degradation (Fig. 3). In 2005, the SBNMS archaeologists reported that the deck planking on the freight deck and its deck beams were "all articulated and nearly intact" (SBNMS 2005), but they also noted that the windlass (viewed on the freight deck in 2003) had fallen into the chain locker, likely as "a result of the deck giving way under the machinery's weight" (SBNMS 2005). During the 2019-2020 mission, the windlass was unable to be documented because there were new net intrusions creating hazardous obstructions for the ROV.
From 2010 to 2021, considerable structural loss and change occurred on Portland's freight deck (Figs. 3 and 10). Nearly 70% of the deck planking was absent, and many of the deck beams had been broken or become disarticulated. Approximately 15.6 m (46.8 ft) forward of the funnels, there was a 21.9 m 2 (70 ft 2 ) opening where the deck beams  Non-metric multidimensional scaling plots showing differences in community structure across years for the Portland's port bow (a), freight deck (b), and fantail stern (c). Each point represents one frame grab sample, and the distance between points indicates their level of dissimilarity in community composition between them. Clustering of points by year indicates differences in community structure across time As Fig. 10 illustrates, there was a steep decline in the percent of available habitat between 2010 and 2021 on the freight deck of Portland. Approximately 68% of the deck had either decking or deck beams-a wooden surface for settlement of marine organisms-in 2003 and 2010. In 2021, the percentage of the deck that contained wood was only 40%. This degradation and reduction in the available habitat corresponded to impacts on the biological community.
On the freight deck, there was a gradual decline in the abundance of the branching sponge Haliclona oculata 2003-2020 (Fig. 11). Three species-the round sponge Polymastia robusta, the hydroid Ectopleura crocea, and the tunicate Molgula sp.

Fantail
Another significant discovery was the collapse of the stern fantail. Figure 4 shows Portland's elegant elliptical fantail stern standing proud off the seafloor in the 2002 and 2010 sonar images but absent in the 2021 side-scan. In 2019-2020, we observed that the fantail along with railing had been disarticulated from the sternpost and rudder and now were spread on the seafloor around the stern. Gillnets and associated lines were wrapped and entangled in the collapsed fantail, the rudder, and hull structure on the starboard side. The collapse of the fantail was approximately 33 m from the walking beam engine. The area of collapse was approximately 35.89 m square. The 3D  model clearly shows gill nets and ghost gear wrapping around the port and starboard sides of the ship's stern as well as around the rudder (Fig. 12).
Only two biological sampling years were available for Portland's fantail stern: 2009 and 2020. In 2009, the community was entirely dominated by M. senile, but in 2020, the same structure was completely covered by Molgula sp. (Fig. 13).

Discussion
During its wrecking event, Portland lost its superstructure and settled in a muddy basin. This is a prime area for anthropogenic activities such as gill netting and trawling, which became the more profitable ways of fishing at the turn of the twentieth century. Unfortunately, these fishing methods are highly destructive to the fragile remains of the shipwreck. Early archaeological surveys noted the ubiquitous presence of ghost gear around and throughout the shipwreck, impeding investigations, but SBNMS staff were able to provide baseline characterizations of site formation for us to compare our data sets with. It is clear that the Portland shipwreck site has been continually impacted by c-transforms and that its degradation has increased significantly between 2010 and 2019. Accelerated degradation of the Portland between 2009 and 2019 could potentially be related to changes in fishing pressure. A major change in fishing regulations in 2010, the implementation of sector management, coincided with increased net entanglements on the Portland; however, it is unclear whether this regulatory change increased fishing effort in the area around the shipwreck (ONMS 2020). More recently, there has been increased activity in the scallop fishery in SBNMS since 2017 (ONMS 2020). This fishery involves dredging, which can be very damaging to seafloor communities and maritime heritage structures. While this study cannot demonstrate a definitive cause of Portland's degradation between 2010 and 2019, it is clear that the site formations have had impacts on the biological community at the site.

Port Bow
The major site formation process on the port bow was the entanglement of a fishing net between 2010 and 2019. This flexible net waved in the water column, providing a hydrodynamically different habitat than the solid hull underneath. There are substantial differences in invertebrate communities on substrata that are fixed versus mobile (i.e., waving in the water column) (Perkol-Finkel et al. 2008). The fishing net had much lower abundance of invertebrates overall than the hull of the Portland prior to its entanglement. Aside from microhabitat preference, the growth form of organisms might impact their survival on a shipwreck affected by fishing. Both species that declined on the port bow-M. senile and A. infundibuliformis-are three-dimensional, growing upward and outward from their substrata. In contrast, the one species that increased on the port bow-H. panicea-is an encrusting sponge that grows flat to the surface underneath. Fishing nets become entangled on shipwrecks because fishers attempt to drag their nets as close to the shipwreck as possible without entanglement. Sheering by fishing nets over a decade could explain the replacement of three-dimensional species with a flat, encrusting species on Portland's port bow because H. panicea was less susceptible to being scraped off.

Freight Deck
In 2002, when the freight deck was more-or-less intact, the horizontal surface likely experienced laminar flow and a strong boundary layer of slower-velocity water. Suspensionfeeding species would need to seek elevation out of the boundary layer to feed, for example by settling on top of artifacts or by having branching structures (Vogel 1996). The branching sponge Haliclona oculata was very common on the freight deck and was particularly abundant on top of artifacts such as the forward winch. As the wood fragmented, the flow regime and available microhabitats changed, leading to a decline in H. oculata. The freight deck collapse transformed a more-or-less horizontal surface into a depression with slanted vertical sides and protruding deck beams ringing the edge. These surfaces were more conducive to species including P. robusta and M. lingua, which prefer vertically oriented surfaces, and E. crocea, which we have observed elsewhere on protruding structures (Meyer-Kaiser et al. 2022a).
It is unclear whether the freight deck collapse can be attributed to c-or n-transforms, or a combination of both. Conceivably, c-transforms like fishing activities and associated damage (i.e. trawling impact and net entanglement) around the bow could have destabilized the planking and deck beams, leading to fragmentation and eventual collapse. Additionally, there are multiple species of shipworms (bivalves in family Teredinidae) and microbial communities (bacteria, archaea, fungi) in Massachusetts that could have degraded the freight deck's wooden structure (Turner 1966).
If the degradation was primarily the result of these n-transforms, then it is unclear why other parts of the wooden shipwreck did not also collapse. A possibility is that the freight deck was one of the few extant horizontal features on the shipwreck that supported several other heavy artifacts (for example, the steering apparatus and anchor windlass) that stressed the degraded deck beams. Further, while it is uncertain why the most extensive change in the freight deck integrity occurred 110-120 years after sinking, many cultural heritage sites undergo swift degradation followed by a state of environmental equilibrium, followed by another period of degradation (Muckelroy 1978). It is possible that shipworm or microbial activity over a century degraded the freight deck enough that it finally gave way.
Finally, a combination of c-and n-transforms may have played combined roles to varying degrees. Shipworm or microbial activity may have weakened the planking and beams. Then, fishing activities may have impacted the structure and served as a catalyst to destabilize the status quo and hastening the deck's collapse. We are, however, unable to draw a clear conclusion about the cause of the freight deck collapse, but cannot attribute the change solely to anthropogenic influence.

Fantail
The fantail saw the greatest structural loss, and we can be more certain that c-transforms from fishing activities had a significant impact. Fantail railing and timbers were completely detached and disarticulated from the wreck. Remnants were strewn around the stern. There were gillnets entangled around the rudder and collapsed fantail. The fantail stern saw a complete replacement of M. senile with Molgula sp. after its collapse. Metridium senile is a species that preferentially settles on elevated structures, and it also occurs in high abundance on Portland's A-frame (Meyer-Kaiser et al. 2022a). Elevation facilitates feeding in this species, which is a zooplankton predator (Sebens and Koehl 1984). Molgula sp., by contrast, may not require as much flow. Tunicates can actively pump water through their pharyngeal baskets, so feeding in this species does not depend on elevation. Molgula sp. is common on low-lying boulder reefs in SBNMS and elsewhere.

Changes in Pelagic Biodiversity Over Time
The changes we observed in fish communities cannot be directly attributed to structural changes in the Portland shipwreck and are more likely reflective of broader-scale changes across the region. Auster and Conroy (2019) found that in the decade 2009-2019, there was a net increase in Acadian redfish across Stellwagen Bank National Marine Sanctuary, which is reflected in our data as well. The reasons behind this increase could be related to climate change, overfishing of competitive species, or population recruitment dynamics (Auster and Conroy 2019). The complete dominance of Atlantic cod in 2006, on the other hand, is more likely related to stochastic factors. Because ROV video was only recorded over a limited time-period, the occurrence of a large group of cod near the Portland could have been a temporary phenomenon.
Acadian redfish use heterogenous reef habitats throughout their life-histories, including as nurseries (Auster et al. 2003). Other fish species that we observed near the Portland shipwreck prefer rough seafloor habitats such as gravel and boulder patches, so it is intuitive that they 1 3 should utilize the structure provided by a shipwreck. However, since these species primarily use low-lying gravel and boulder reefs, they may be relatively unaffected by the site formation processes described here. For example, cusk and wolffish (Anarhichas lupus) were primarily observed in and amongst the debris field surrounding the Portland and in low-lying regions of the shipwreck. Degradation of Portland's tall vertical structures such as the freight deck and fantail would leave these low-lying structures unchanged. In fact, piles of ballast rocks or cargo from older shipwrecks with degraded hulls (known as ballast reefs; Meyer-Kaiser and Mires 2022) form the closest analogue to natural hard-bottom habitats (Meyer-Kaiser et al. 2022b). Both cusk and wolffish have declined in the northwestern Atlantic in recent decades, leading to questions about habitat availability and continuity (Hare et al. 2012;Novaczek et al. 2017). Even in an advanced state of degradation, shipwrecks can provide essential habitat for these at-risk species.

Changes to Invertebrate Community Due to Changes in Ship Structure
The changes we observed in invertebrate community composition speak to complex interactions between shipwreck structure and the associated biological communities. Sessile invertebrates, such as the sponges and anemones we observed on the Portland, preferentially settle in certain microhabitats that offer protection from predators or hydrodynamic advantages for feeding (Maldonado and Uriz 1998). In fact, larvae exhibit strong preferences for particular hydrodynamic environments during settlement (Mullineaux and Butman 1990;Mullineaux and Garland 1993). The range and availability of microhabitats can influence invertebrate community structure.

Cautions
Unfortunately, we cannot exclude the possibility the changes in community structure that we observed on the Portland shipwreck were influenced by external factors such as water temperature changes or regional population dynamics. However, these factors would be expected to act universally across the shipwreck, and most of the changes in abundance we observed for species varied between different regions of the shipwreck. For two species-A. infundibuliformis and Molgula sp.-the same general trends were observed for the port bow and the freight deck, indicating they may be the result of external factors. Axinella infundibuliformis is a cold-water sponge species that is at the southern end of its geographic range in SBNMS. Bottom waters in the Gulf of Maine are rapidly warming, so we cannot exclude the possibility that this species declined as a result of regional temperature change rather than structural change in the shipwreck. For Molgula sp., the high abundances observed in 2019-2020 could be the result of differences in sampling methods: the ultra-high-definition video recorded in these years allowed better visualization of the small brown tunicates compared to previous years. Given these uncertainties, we have interpreted our results cautiously and focused on obvious trends that have some plausible mechanism related to shipwreck structure.

Conclusion
Our study on site formation processes on the shipwreck Portland shows very clearly that anthropogenic fishing activity has a huge impact on the structure of the shipwreck and subsequently on the biological community. Sheer from fishing nets being dragged over the wreck and the entanglement of a trawl net altered the invertebrate community on the port bow. Similarly, fishing damage to the fantail stern caused a complete change in the species composition of invertebrates there. Not every site formation to Portland over the last 20 years is cultural, though. The collapse of the freight deck and changes in the fish community can be attributed to natural causes. While changes in the invertebrate communities can be tied to changes in shipwreck structure, changes in community composition are not necessarily negative. The species that had changes in abundance in this study are all sessile, suspension-feeding invertebrates. These taxa fill similar ecological roles by feeding on plankton and detritus and serving as prey for higher trophic levels. Functional redundancy between invertebrate species means that changes in their relative abundances are unlikely to have broader ecosystem-level implications. The primary implication of fishing activity on and near the Portland shipwreck is damage to the structure of this irreplaceable cultural resource. Studies in Maritime Heritage Ecology such as this one reveal important interactions between biological and archaeological processes, including n-and c-transforms, and can inform critical management decisions.