Data From: Shortened food chain length in a fished versus unfished coral reef

Direct exploitation through fishing is driving dramatic declines in wildlife populations in ocean environments, particularly for predatory and large-bodied taxa. Despite wide recognition of this pattern and well-established consequences of such trophic downgrading on ecosystem function, there have been few empirical studies examining the effects of fishing on whole system trophic architecture. Understanding these kinds of structural impacts is especially important in coral reef ecosystems - often heavily fished and facing multiple stressors. Given often high dietary flexibility and numerous functional redundancies, especially in diverse ecosystems such as coral reefs, it is important to establish if web architecture is strongly impacted by fishing pressure, or if it might be resilient, at least to moderate intensity pressure. To examine this question, we used a combination of bulk and compound-specific stable isotope analyses measured across a range of predatory and low trophic level consumers between two coral reef ecosystems that differed with respect to fishing pressure but otherwise remained largely similar. We find that even in a high diversity system with relatively modest fishing pressure, there are strong reductions in the trophic position of the three highest trophic position consumers examined in the fished system but no effects on the trophic position of lower-level consumers. We see no evidence that this shortening of these affected food webs is being driven by changes in basal resource consumption, e.g., through changes in spatial location of foraging by consumers. Instead, this likely reflects internal changes in food web architecture suggesting that even in diverse systems, and with relatively modest pressure, human harvest causes significant compressions in food chain length. This observed shortening of these food webs may have many important emergent ecological consequences for the functioning of ecosystems impacted by fishing or hunting. Such important structural shifts may be widespread but unnoticed by traditional surveys. This insight may also be useful for applied ecosystems managers grappling with choices about the relative importance of protection for remote and pristine areas and the value of strict no-take areas to protect not just the raw constituents of systems affected by fishing and hunting but also the health and functionality of whole systems.

Methods
More details of data collection are provided in the Ecological Applications Publication "Shortened food chain length in a fished versus unfished coral reef" (Young et al., 2024). This work was conducted in the northern line islands in the central Pacific between 2006 and 2009. All samples were collected from Palmyra Atoll National Wildlife Refuge (5° 53’ N, 162° 05’W), a US outlying island, and Tabuaeran (3° 51’ N,159 21 W), part of the nation of Kiribati (Fig 1A). These two tropical coral reef atolls are situated in marine regions with similar sea surface temperatures (27.9° C vs 27.5° C respectively) and have a similar coral cover (20.4% and 19.5% respectively, Sandin et al. 2008). Palmyra is largely uninhabited (population of 6 to 25) and as a marine protected area, fishing is almost entirely prohibited within 50 nautical miles. Tabuaeran at the time of this survey had a population density of approximately > 60 humans per kilometer of the reef (2500 total) and allowed both commercial and artisanal fishing.

Average biomass
Fish communities were characterized at nine locations at Palmyra and five locations at Tabuaeran. These sites were distributed approximately 2 km apart from each other at random coordinates in shallow water (3-12m) with replicate surveys conducted. Fish surveys were conducted four times each at Tabuaeran (along the western and southern coastlines; March - April 2007) and seven times each at Palmyra (both northern and southern coastlines; June - August 2006). Fish communities at both sites were surveyed using belt transect surveys conducted in the daytime. At each site, a fish survey was composed of four belt transects, their dimensions tailored to the size (total length; TL) of the fish being surveyed: fish of ≥50 cm TL were counted along a 50×8 m transect; 30–49 cm TL fish along a 50×4 m transect; 15–29 cm TL fish along a 50×4 m transect; and fish of <15 cm TL were surveyed along a 25×2 m transect. Within each transect, a pair of observers were responsible for identifying, counting, and estimating the total length of each fish. The same pair of divers were responsible for all surveys across both atolls. Estimation of fish biomass was then derived from the survey data, using length-weight conversion constants sourced from FishBase (Froese and Pauly 2010) or other scholarly publications. Further details are provided in (McCauley et al. 2012a). Fish were placed into one of four basic trophic classifications for subsequent analyses: predator, invertivore, herbivore, and planktivore. Trophic classifications were made based on the dominant diet type as assessed via a review of diet information in Fishbase (Froese and Pauly 2010), as well as a review of literature on species diet for these species from this region of the Pacific.  

Benthic cover
The benthic cover was surveyed by estimating percent cover by cover type across a series of 10 gridded quadrats at each site; each 1 m2 quadrat was distributed at 5 m intervals along a 50 m transect. Before analysis, all cover data was pooled per site and binned in general categories: algae (both turf and macroalgae), living coral, dead substrate (sand, rock, rubble, and dead coral), and other (which predominantly included crustose coralline algae and other invertebrates). Benthic data from the 10 quadrats was then pooled by the site for all subsequent analyses. 

Data processing and pooling are described separately for each sheet of data. However, in general, isotope data is unaggregated (presented at the level of the individual) while fish and benthic data are pooled at the transect level (e.g. average fish per transect per day or average cover per transect across all blocks).