Ecological legacy effects refer to preexisting anthropogenically induced conditions in ecosystem structure and the environment that continue to affect processes in the present day (Silliman et al., 2018). Temporal scales of legacy effects range from decades to hundreds or even thousands of years, and spatial scales encompass whole ecosystems, catchments, or biogenic regions. Because of the large spatial and long temporal scales of antecedent effects on ecosystems, their study can be somewhat elusive for want of effective controls as bases for comparison in both past and present-day systems. The concept of shifting baselines of current ecological perspectives on food web structure is a key example of this problem (Dayton et al., 1998; Pauly, 1995). Successful resolution of the “ghosts of past ecological relationships” relies on effective ways of gleaning ecological information from prehistoric faunal material and historical accounts as well as comparative studies of degraded and “pristine” ecological systems in the present day (Connell, 1980; Erlandson et al., 2007).
Information gleaned from zooarchaeological collections has provided key information on the spatial and temporal variability of the ecological and environmental systems in prehistoric times in a variety of systems, allowing a basis for comparison with historic and current ecosystems (Jackson, 2001; Simenstad et al., 1978; Smith, 2013; Wing, Durante, et al., 2022; Wing, Shears, et al., 2022; Wing & Wing, 1995, 2001). A complementary solution is to contrast different present-day ecological systems with marked differences in their histories of disturbance, land conversion, exploitation, or other anthropogenic influences (e.g., Schlieman et al., 2022; Udy, Wing, O'Connell-Milne, et al., 2019). The approach relies on processes within contrasted regions to be relatively independent and isolated from broad-scale degradation, as well as being comparable in terms of key ecological processes at play (Udy, Wing, Jowett, et al., 2019). In marine systems, the study of “intact” ecosystems has been vital to achieving this goal by providing key insights into the importance of land–sea connections, predator-mediated ecological processes, animal-mediated nutrient fluxes and, importantly, the interactions among cumulative long-term and broad-scale stressors.
No-take marine reserves have been a useful tool for highlighting localized effects in kelp forest habitats in this context (Babcock et al., 2010; Shears & Babcock, 2002), but when the reserves are small and isolated rather than parts of a larger connected network, they do not generally provide complete insights into processes acting across the regional scale of reproductive source–sink and metapopulation dynamics of key species (Jack & Wing, 2013; Quinn et al., 1993; Wing & Jack, 2013) or large-scale patterns in diffuse effects from land-based stressors (Schiel, 2013), large-scale disturbances, and climate change (Schiel et al., 2014; Smale et al., 2019; Thomsen et al., 2019). Kelp forests provide essential three-dimensional biogenic habitats and sources of organic matter that support a wide diversity of species (Mann, 1973; Schiel & Foster, 2015). As a consequence, the loss or decline of habitat-forming algae through disturbance or environmental degradation can induce long-lasting, cascading effects on diversity in kelp forest communities (Schiel et al., 2019, 2021; Smale & Moore, 2017).
Dating to prehistory, the consequences of land conversion for agriculture and settlement as well as marine hunting and gathering in the coastal zone have induced profound changes to the environmental conditions and the ecosystem structure in many parts of the world. New Zealand was first settled with the arrival of Māori during the little ice age ca. 1250 ad, followed by widespread colonization by Europeans in the late 1700s (Smith, 2020). During the early period, extensive land clearance by fire-assisted hunting and slash and burn agricultural practices converted extensive areas of native forests into grassland or scrub (McWethy et al., 2009). With the arrival of Europeans, deforestation intensified and vast areas of native forest were converted to grasslands for agriculture, urban development, forestry, and mining (Ewers et al., 2006). By the early 1900s, relatively small areas were still covered by virgin native forests and modified landscapes dominated the eastern and northern coasts of the South Island and much of the nonmountainous regions of the North Island (Proce et al., 2019).
Clear-felling of native forests in particular has resulted in increased erosion, and large inputs of fine sediments and debris to coastal marine habitats that have resulted in direct changes in the distribution of habitat-forming organisms, such as corals in tropical environments (e.g., Halpern et al., 2013), kelp forests, and other biogenic habitat providers in temperate regions (Foster & Schiel, 2010; Krumhansl et al., 2016; Schiel & Foster, 2015). In kelp forests, increases in turbidity can result in a shoaling of the compensation point and contraction of habitat suitable for the growth of sporophytes (Blain et al., 2021; Shepherd et al., 2009; Tait et al., 2014; Tait & Schiel, 2011). In addition, spectral shifts in photosynthetically active radiation (PAR) and smothering by fine sediments can reduce the viability of gametophyte stages of the life cycle, further reducing the habitat suitable for, and recovery potential of, kelp forests (Devinny & Volse, 1978; Schiel & Foster, 2006; Tait, 2019). These effects can be nonlinear where rocky reef habitats form a bench structure across depth, resulting in rapid contraction of kelp forest habitats as fine sediments accumulate at depth and extinction of PAR in the water column increases. Both coastal darkening and smothering by fine sediments are particularly important drivers of long-term contractions of kelp forests from deeper reef habitats (e.g., Blain et al., 2021).
Hunting, fishing, and gathering in marine coastal systems have a history that spans millennia and has resulted in long-term shifts in the structure of marine food webs globally (Estes et al., 2011; McCauley et al., 2015; Wing, Durante, et al., 2022; Wing, Shears, et al., 2022). Removal of large, top predatory species in particular has resulted in simplified food web structure and loss of some trophic interactions including the regulation of sea urchin populations (Dayton et al., 1995; Estes et al., 2004; Thrush & Dayton, 2010). Losses of large predatory fishes from the coastal zone in New Zealand over the last century have resulted in distinct changes in the structure of food webs with unknown effects on grazer–kelp dynamics within kelp forests (Durante et al., 2022; Wing, Durante, et al., 2022; Wing, Shears, et al., 2022). For example, the Hāpuku (Polyprion oxygeneios: Polyprionidae, wreckfishes) is a large-bodied predatory fish that was formerly an integral part of coastal kelp forest food webs but has been all but lost from the nearshore in modern times, a legacy of overexploitation on coastal rocky reefs (Graham, 1953; Wing, Durante, et al., 2022; Wing, Shears, et al., 2022). In addition, widespread exploitation, with the expansion of coastal fisheries in the 1930s, of important sea urchin predators such as snapper (Pagrus auratus: Sparidae, sea breams), the red rock lobster (Jasus edwardsii), packhorse lobster (Sagmariasus verraeuxi), and blue cod (Parapercis colias: Pinguipedidae, weevers) has resulted in sharp declines in abundance and truncation of size distributions (Durante et al., 2020) both of which decrease the effectiveness of these species as sea urchin predators (Andrew & Macdiarmid, 1991; Babcock et al., 1999; Shears & Babcock, 2002).
It has been widely demonstrated that no-take marine reserves support more abundant and old-growth size distributions of key sea urchin predators (Babcock et al., 1999; Jack & Wing, 2010; Pande et al., 2008). Globally, observations of the strong interaction between grazing sea urchins and the persistence of kelp forests have provided the basis for explaining state changes in kelp forest habitats from productive algal stands to sea urchin barrens (Filbee-Dexter & Scheibling, 2014; Tegner & Dayton, 1991). Nevertheless, state changes typically have multiple causalities in ecological systems, and the predator–sea urchin–kelp interaction chain alone is often insufficient to explain all of the dynamics observed in these systems as they degrade (Dayton et al., 1998; Foster & Schiel, 2010; Sala et al., 1998; Shears et al., 2008). For example, the legacy of overexploitation in kelp forest systems has also removed dense concentrations of abalone in many areas around the world (Hobday et al., 2001) including in New Zealand (Wing et al., 2015). There is some evidence of competition for space between sea urchins (Evechinus chloroticus) and abalone (Haliotis iris), suggesting that in the absence of exploitation, abalone can effectively reinvade and maintain patches free of sea urchin grazing in the shallow wave-washed zone, thereby preventing overgrazing of kelps by sea urchins (Wing et al., 2015). Each of these patterns indicates possible long-term legacy effects of changes in the trophic structure of kelp forest systems with unknown direct and indirect effects on community, and grazer–kelp dynamics.
The likely interactions and feedbacks between the environmental degradation from both chronic and large-scale episodic land-based stressors (nutrient pollution and sedimentation) that undermine the basic environmental infrastructure of macroalgal assemblages (Schiel et al., 2021), and the effects of exploitation on food-web architecture (trophic downgrading and simplification of food webs), highlight the importance of understanding cumulative effects on the legacy of ecological processes in kelp forest systems (Castorani et al., 2021). Observations of both top–down and bottom–up controls leading to alternate stable states in kelp forests (Falkenberg et al., 2013; Russell et al., 2009; Russell & Connell, 2012) have highlighted likely important, but largely unresolved, interactions between land-based degradation of light, nutrient, and substratum conditions on rocky reefs and the effects of exploitation on food web architecture and interaction webs within the system (Filbee-Dexter & Scheibling, 2014). Effects of the dominant stressors are likely differently distributed across depths on rocky reefs. For example, increases in sedimentation and declines in light penetration primarily affect the maximum depths at which kelp can grow and productivity gradients across depth, while increases in grazing pressure may be more concentrated in shallow regions of reefs where grazers tend to aggregate.
To better understand these potentially important interactions, we examined depth-specific patterns in the association of the common kelp (Ecklonia radiata) and New Zealand sea urchin (Evechinus chloroticus) across a range of different environments in no-take marine reserves and fully fished areas in three biogeographic regions with contrasting histories of land conversion and fisheries exploitation. Satellite-derived total suspended solids (TSS) were used as a proxy for the effects of land-based sedimentation, estimates of the regional densities of Evechinus were used as a proxy for the potential magnitude of larval supply of new recruits from the regional population network, and marine reserve effects on Evechinus density were used as a proxy for the importance of predation in regulating grazing pressure by adult Evechinus. We expected that in areas where suspended sediment loads were high, Ecklonia density would decline sharply with depth, independent of Evechinus abundance. Where land-based influences dominated, we expected that marine reserve effects would be relatively small at depth, indicative of regionally diffuse effects of dominant stressors. We expected that in regions where fishing pressure has been historically intense, marine reserve effects would be relatively large, particularly in the more shallow depth strata, as evidenced by reduced sea urchin density and greater kelp abundance in reserves. Finally, we expected that where land-based stressors were minimal and fishing pressure was less intense, marine reserve effects would be small and Ecklonia and Evechinus would be distributed across depths with a deep refuge from grazing for kelp facilitated by relatively deep penetration of PAR, and low suspended sediments.
Ecklonia forms dense subtidal beds in each of the three regions but is most abundant in the northern region, where extensive monospecific stands dominate many subtidal platforms. Accordingly, absolute abundances of Ecklonia beds naturally vary by region with local conditions of light penetration, temperature, and wave action, but their relative responses to diffuse or localized anthropogenic stressors likely reflect the long-term legacy of stressors at the regional level. The results provide important evidence for assessing the relative importance of and interactions between the legacies of land clearance and trophic downgrading in New Zealand rocky reef habitats across the depth and among the major regions in a cumulative stressors framework to inform mitigation measures required for effective ecosystem-based management.
MATERIALS AND METHODS Study regionsThe primary biological and the remotely sensed environmental data for the study were collected from three important biogeographic regions of New Zealand (Figure 1). The northern region was centered around Leigh and Tawharanui, and had several sites that were sampled within and outside of no-take marine reserves. The central region encompassed the Marlborough Sounds and Tasman Bay, with sheltered sites in the Sounds and sites under the influence of Cook Strait within and outside of no-take marine reserves. The southern region comprised sites within nine fjords in the Fiordland region in the southwest of the South Island of New Zealand both within and outside of nine no-take marine reserves. All regions had two or more well-established no-take marine reserves, allowing for an orthogonal sampling design of multiple reserve and fished kelp forest sites. Satellite-derived proxies for the subtidal environmental conditions were available for each region.
Enlarge this image.FIGURE 1. Charts of New Zealand with locations of primary marine reserve (red circles) and reference (blue circles) areas with multiple sites in each of the following three regions: (a) northern, (b) central, and (c) southern. Terrestrial land cover categories were derived from the Land Information New Zealand database.
A time series of remotely sensed water quality data were available for study sites at Leigh and Tawharanui (northern region), Long Island and Tonga Island (central region), and Milford Sound to Breaksea Sound in the southern Fiordland region (Figure 1). Data were extracted from “NIWA-SCENZ” (Pinkerton et al., 2021), which is an open-source (
The full methodology for calculation of satellite-derived products can be found at
ADET measures the amount of light absorption in the blue portion of the spectrum (wavelength = 443 nm) by detrital material. The material can include both colored dissolved organic matter (CDOM) and suspended sediment covered by an organic film. The CDOM or “yellow substance” is a complex mixture of humic and fulvic acids formed during the breakdown of organic matter in soil or water. Elevated ADET can result from CDOM and suspended sediment discharged from rivers, and can be a useful indicator of freshwater input into the coastal zone. The total detrital absorption coefficient at 443 nm (in per meter), including that due to CDOM and particulate detrital absorption, was estimated using the quasianalytic algorithm (QAA) (Lee et al., 2002, 2005) as described in the CHL methodology (Pinkerton et al., 2021).
The CHL product indicates the amount of phytoplankton in the water and is indicative of biomass and pelagic primary production. Surface measurements of chl a are indicative of water-column integrated phytoplankton abundance and, at broad time and space scales, ocean productivity (Aiken et al., 1992; Campbell et al., 2002). Suspended sediment and CDOM make it scientifically challenging to estimate chl a in coastal waters from satellite data. Management Unit of the North Sea Mathematical Models atmospheric correction processing (Ruddick et al., 2000) was used with α = 1.945, implemented in SeaDAS, v7.2, as this gave the best results when tested against in situ data around New Zealand (Pinkerton et al., 2019), as elsewhere (e.g., Ody et al., 2016). QAA (Lee et al., 2002, 2005) was used to estimate particulate backscatter at 555 nm [BBP(5Black 55)] and phytoplankton absorption at 488 [aph(488)]. QAA uses an inversion of the spectral remote-sensing reflectance to derive absorption and backscatter coefficients of the water column. Phytoplankton absorption was converted to an estimate of chl a using the chl-specific absorption coefficient, aph*(488). The value of aph*(488) can vary seasonally and spatially, related to different phytoplankton species (varying cell physiology and pigments), different phytoplankton cell sizes, and the light environment (Kirk, 2011). To obtain CHL, average values found for oceanic phytoplankton (Bissett et al., 1997; Bricaud et al., 1995) and measurements in the lower reaches of New Zealand rivers and estuaries (Pinkerton, 2017) were used (Pinkerton et al., 2021). The QAA-chl a and the MODIS-default chl a product (NASA, 2018a) were blended using a logistic-scaling of BBP(555) (Pinkerton et al., 2018).
HVIS, a measure of underwater visibility, is defined as the horizontal black disk visibility (i.e., the horizontal distance at which a black disk can be distinguished from the background when viewed underwater by an observer at the same depth). HVIS is a better index of visibility than the Secchi depth because it is insensitive to ambient lighting (e.g., whether direct or diffuse sunlight) and change in light with depth (Davies-Colley, 1988; Zaneveld & Pegau, 2003). HVIS was estimated as 4.8/c(550) (Davies-Colley, 1988). QAA (Lee et al., 2002, 2005) was used to estimate absorption and back-scattering coefficients at 550 nm (as described in the CHL methodology above), and a back-scattering/total-scattering ratio of 0.0312 was used based on New Zealand's National River Water Quality Network database (Pinkerton et al., 2021; N = 236; r2 = 0.816).
The SST time series were obtained from the MODIS-Aqua measurements using the SeaDAS, v7.2 default “sst” product, which was derived from measurements of long-wave (11–12 μm) thermal radiation (NASA, 2018b). The SST products at 1 km were subsampled to 500 m to improve the resolution of the narrow channels over time using bilinear interpolation. The accuracy of SST is likely to be high. Validation comes from a comparison between MODIS-Aqua SST and Optimum Interpolation Sea Surface Temperature (Reynolds et al., 2002) for the New Zealand coast (Pinkerton et al., 2019) (n = 256,687, r2 = 0.972).
The TSS concentration is an estimate of the total gravimetric concentration of particulate material in the upper water column. It includes both algal (phytoplankton) and nonalgal particulates, the latter including inorganic matter (suspended sediment) and nonalgal detrital particulate material. The TSS product was derived from a nonlinear scaling of particulate backscatter at 555 nm (see BBP) and defined as follows: TSS = 58.00 × ([0.8513 × BBP]0.8701). The applied relationship was based on a standardized major axis regression between measurements of backscatter at 660 nm (ECO Triplet, WET Labs, USA) and gravimetric measurements of TSS in many areas around New Zealand (Pinkerton et al., 2021; n = 367, r2 = 0.928, p < 0.001).
Subtidal surveys ofDiving surveys using randomly placed quadrats were used to quantify the density and depth distribution, between mean low tide and 15 m, of the common kelp Ecklonia radiata and the sea urchin Evechinus chloroticus. Subtidal surveys were done at multiple paired sites in marine reserves and areas open to fishing in the central region of the Marlborough Sounds and Tasman Bay during January 2017 and February 2018 (18 sites, n = 286) from the vessel RV Polaris II. Surveys of no-take marine reserves and areas open to fishing in the southern region of Fiordland were done among nine fjords during February 2006 (25 sites, n = 935), February 2010 (21 sites, n = 785), November 2015 (9 sites, n = 158), and May 2017 (6 sites, n = 82), (total 31 sites, n = 1960) from the vessels MV Renown, MV Southern Winds, RV Polaris II, and RV Typhoon. Dive surveys in the northern region were done in 2019 with multiple sites inside and outside of the Leigh Marine Reserve and the Tawharanui Reserve (18 sites, n = 248). Subtidal survey data were stratified into shallow (0–5 m), mid (6–10 m), and deep (11–15 m) depth bins for statistical comparisons among depth strata and regions.
Statistical analysesWe examined the relationship between quadrat counts of Evechinus and Ecklonia across regions to test for differences in the associations between species at the 1-m2 spatial scale. Because these datasets at the quadrat level were zero-inflated and the distribution of counts differed by region, we fitted generalized additive models for location, scale, and shape using the GAMLSS package in R to test for differences in localized associations between Evechinus and Ecklonia. Models were run using the negative binomial distribution family. A key feature of GAMLSS models is that data can be separated into the mean of nonzeros (μ) similar to a conventional generalized linear model, a scale parameter (σ) to test differences in variance among factors, and the probability of zero (ν) among factors. In this application, the probability of finding zero Ecklonia (ν) was particularly useful because it indicated a local gap in the kelp forest habitat.
Generalized linear mixed models (GLMM) were then used to test for differences in the distributions of Evechinus and Ecklonia by REGION (three levels, fixed) among DEPTH ZONES (three levels, fixed), marine RESERVE status (two levels, fixed) with DISTANCE from the wave-exposed coast (continuous) included as a covariate to account for the primary natural environmental gradients from the wave-exposed coast into the sounds and fjords (Wing et al., 2007). Using the above model framework, we calculated marine reserve effect sizes from the GLMM for Evechinus and Ecklonia by depth strata within each region.
A model selection approach was used to investigate the relative importance of (1) TSS (in grams per cubic meter), (2) regional density of Evechinus chloroticus (number per square meter), and (3) marine reserve effects on the density of Evechinus chloroticus (number per square meter) on Ecklonia density within each depth zone across all regions. Site-averaged values of density of Ecklonia for each marine reserve and reference site (n = 63) stratified by depth zone (0–5, 6–10, and 11–15 m) were modeled using multiple GLMMs consisting of each combination of the three variables, with DISTANCE included as a covariate as above, and corrected Akaike information criterion (AICc) values, as well as factor weightings, were calculated. The analysis comprised data from nine marine reserves in the southern region and two marine reserves in each of the northern and the central regions to provide information on the relative weight of the three factors for a statistical explanation of variance in Ecklonia density across the range of observed variability in the three factors among regions. The three factors used in AICc were derived from the most parsimonious effects model based on an analysis of combinations of all potential measured effects, both environmental and biological.
RESULTS Differences in physical conditions among study regionsRegional comparisons from 1 June 2002 to 1 July 2021 for ADET, BBP, CHL, black disk HVIS distance, KPAR, vertical visibility of a black and white SEC, SST, and TSS revealed distinct differences in the environmental conditions among regions (Figure 2; Appendix S1: Table S1). Importantly, BBP and TSS were particularly high in the central region relative to conditions in the northern and southern regions. As a corollary, HVIS was generally confined to 10 m in the Marlborough Sounds relative to conditions in the northern and Fiordland regions, where HVIS encompassed all three depth strata of the subtidal surveys. As expected, SST was relatively high in the northern region, while CHL and KPAR were not detectibly different among the regions using the available time series (Figure 2; Appendix S1: Table S1). The primary environmental results were relatively consistent over time (Figure 2). The northern region had warmer water and greater seasonal fluctuations in SST than the central and southern regions. The central region had a far higher mean average of TSS although there were very pronounced spikes in the northern region, consistent with heavy rain-induced sedimentation events. The central region had much lower HVIS than the other two regions over 20 years, except during four distinct spikes of poor visibility in the north (Figure 2). Importantly, HVIS frequently fell below 10 m in the central region indicating decreased penetration of light to the deepest depth strata included in the analysis of Evechinus and Ecklonia interactions (Figure 2c).
Enlarge this image.FIGURE 2. Time series climatology of (a) sea surface temperature (in degrees Celsius), (b) total suspended solids (in grams per cubic meter), and (c) black disk horizontal visibility distance (in meters) between 1 June 2002 and 1 July 2021 in the three regions from left to right, northern, Marlborough Sounds and Tasman Bay (central), and Fiordland (southern). Each time series is fitted with a fifth-order polynomial regression to visualize temporal changes in mean conditions.
Evechinus and Ecklonia density and their covariance had distinct differences among regions at the 1-m2 spatial scale. For example, counts in the north (Figure 3a) had the greatest densities of Ecklonia of up to 70 individuals/m2, compared to 10–20 individuals/m2 in the central and southern regions (Figure 3b,c). It was interesting to note that even where there were high densities of Evechinus in the north, there could be high densities of Ecklonia at the 1-m2 scale. In the central region, few Ecklonia occurred beyond a density of 5 urchins/m2 and 8 urchins/m2 in Fiordland. Overall, the probabilities of zero Ecklonia in the presence of Evechinus were higher in the Marlborough Sounds and Fiordland (Figure 3b,c; Table 1).
Enlarge this image.FIGURE 3. Coincidence of Evechinus chloroticus and Ecklonia radiata (number per square meter) for (a) Leigh–Tawharanui, (b) the Marlborough Sounds–Tasman Bay, and (c) Fiordland.
TABLE 1 Results of the GAMLSS models for averages of nonzeros and probability of zero: (a)
| Depth zone by region and management zone | Average (no./m2) | SD | N nonzero | N zero | Probability of zero |
| (a) Ecklonia | |||||
| Fiordland | |||||
| MR | |||||
| Shallow | 0.72 | 1.88 | 566 | 442 | 0.78 |
| Mid | 1.11 | 2.16 | 534 | 367 | 0.69 |
| Deep | 0.91 | 1.74 | 520 | 355 | 0.68 |
| Open | |||||
| Shallow | 1.45 | 2.93 | 1087 | 703 | 0.65 |
| Mid | 2.15 | 3.16 | 974 | 481 | 0.49 |
| Deep | 1.97 | 2.91 | 990 | 499 | 0.50 |
| Northern | |||||
| MR | |||||
| Shallow | 45.62 | 27.70 | 47 | 1 | 0.02 |
| Mid | 48.51 | 26.72 | 39 | 0 | 0.00 |
| Deep | 31.61 | 12.35 | 36 | 0 | 0.00 |
| Open | |||||
| Shallow | 8.32 | 23.31 | 50 | 39 | 0.78 |
| Mid | 26.21 | 17.34 | 39 | 4 | 0.10 |
| Deep | 24.97 | 11.44 | 37 | 0 | 0.00 |
| Marlborough | |||||
| MR | |||||
| Shallow | 2.03 | 5.59 | 72 | 60 | 0.83 |
| Mid | 2.48 | 4.90 | 27 | 19 | 0.70 |
| Deep | 0.00 | 0.00 | 29 | 29 | 1.00 |
| Open | |||||
| Shallow | 0.52 | 1.83 | 278 | 243 | 0.87 |
| Mid | 0.26 | 1.07 | 132 | 122 | 0.92 |
| Deep | 0.05 | 0.27 | 162 | 156 | 0.96 |
| (b) Evechinus | |||||
| Fiordland | |||||
| MR | |||||
| Shallow | 0.60 | 1.80 | 566 | 430 | 0.76 |
| Mid | 0.24 | 0.91 | 534 | 463 | 0.87 |
| Deep | 0.09 | 0.37 | 520 | 484 | 0.93 |
| Open | |||||
| Shallow | 1.56 | 3.36 | 1087 | 673 | 0.62 |
| Mid | 0.41 | 1.45 | 974 | 816 | 0.84 |
| Deep | 0.12 | 0.61 | 990 | 924 | 0.93 |
| Northern | |||||
| MR | |||||
| Shallow | 2.21 | 5.43 | 47 | 32 | 0.68 |
| Mid | 0.00 | 0.00 | 39 | 39 | 1.00 |
| Deep | 0.00 | 0.00 | 36 | 36 | 1.00 |
| Open | |||||
| Shallow | 10.32 | 11.20 | 50 | 11 | 0.22 |
| Mid | 2.97 | 5.98 | 39 | 25 | 0.64 |
| Deep | 0.00 | 0.00 | 37 | 37 | 1.00 |
| Marlborough | |||||
| MR | |||||
| Shallow | 3.14 | 3.24 | 72 | 18 | 0.25 |
| Mid | 2.52 | 4.90 | 27 | 13 | 0.48 |
| Deep | 0.48 | 0.83 | 29 | 19 | 0.66 |
| Open | |||||
| Shallow | 4.88 | 9.79 | 278 | 114 | 0.41 |
| Mid | 2.77 | 4.99 | 132 | 69 | 0.52 |
| Deep | 1.47 | 2.42 | 162 | 83 | 0.51 |
Evechinus and Ecklonia had striking differences in density between the reserve and fished locations in the northern and central regions but not in the southern region of Fiordland (Figure 4). The pattern was especially evident in the northern region, where sea urchins dominated the 0–5 m depth stratum and were abundant in the 6–10 m depth stratum at fished sites, but were at very low densities in the no-take marine reserves (Figure 4a). There were high densities of Ecklonia (>15 m−2) across all depth strata in marine reserve sites and lower densities (5–10 m−2) across depth strata at fished sites in the northern region (Figure 4a). In the central region, sea urchins were common across all depth strata (Figure 4b). Compared with patterns in the northern region, there were much lower densities of Ecklonia at all depths, and very low densities in the deepest stratum (Figure 4b). In the central region, Evechinus occurred at 1–2 individuals/m2 in the shallow and intermediate depth strata both inside and outside of the marine reserves (Figure 4b). Densities of Ecklonia in the central region were highest in the shallow and intermediate depth strata at the marine reserve sites, but zero or near zero in the deep stratum both inside and outside of the reserves, even in the absence of high densities of Evechinus (Figure 4b). In the southern region, Evechinus occurred only at low densities (Figure 4c). Densities of Ecklonia were highest at intermediate and deeper depths at both marine reserve and fished sites, and there was no decline in Ecklonia density in the absence of high Evechinus density at the deeper depth stratum (Figure 4c).
Enlarge this image.FIGURE 4. Densities (number per square meter) of Evechinus chloroticus (gray bars) and Ecklonia radiata (black bars) in no-take marine reserves (MR) and open areas (OPEN) among depth strata for (a) Leigh–Tawharanui, (b) the Marlborough Sounds–Tasman Bay, and (c) Fiordland. Error bars are ±1 SE.
Densities of Evechinus averaged across the SITES and DEPTHS were highest (>1 urchin/m2) in the northern region and in the central region and lowest (<0.3 urchins/m2) in the southern region (whole model: F8,2491 = 66.5, r2 = 0.18, p < 0.0001; effects tests: region, F2,2491 = 126.9, p < 0.0001) (Figure 5a). Marine reserve effects for Evechinus density were significantly negative within the whole region and negative within the shallow and mid depth strata in the northern region (<−1 m−2) but non-significant at the deepest depth stratum (whole model: F6,241 = 16.4, r2 = 0.29, p < 0.0001; effects tests: reserve, F1,241 = 35.4, p < 0.0001; reserve × depth zone, F2,241 = 7.79, p = 0.0005) (Figure 5b). Marine reserve effects on Evechinus density were significantly negative within the central region but negative for the shallow depth stratum and nonsignificant for the mid and deep depth strata (whole model: F6,279 = 3.8, r2 = 0.08, p = 0.001; effects tests: reserve, F1,279 = 6.71, p = 0.001; reserve × depth zone, F2,279 = 1.20, p = 0.30) (Figure 5b). In the southern region, marine reserve effects for Evechinus density were significantly negative and negative within the shallow and mid depth strata but nonsignificant at the deepest depth stratum (whole model: F14,1951 = 16.6, r2 = 0.11, p < 0.0001; effects tests: reserve, F1,1951 = 49.9, p < 0.0001, reserve × depth zone, F2,236 = 16.8, p < 0.0001) (Figure 5b). Marine reserve effect on Evechinus density were significantly negative for the northern and central regions but not in the southern region (whole model: F8,2491 = 66.5, r2 = 0.18, p < 0.0001; effects tests: region × reserve, F2,2491 = 32.6, p < 0.0001) (Figure 5c).
Enlarge this image.FIGURE 5. (a) Evechinus chloroticus density (number per square meter) for each of the three regions, (b) marine reserve (MR) effect sizes for Evechinus chloroticus density by depth zone, an asterisk (*) indicates significant effects, and (c) density of Evechinus chloroticus in marine reserves and open areas (OPEN) in each of the three regions. Levels that are not connected by the same letter are significantly different based on Tukey's post hoc tests. Error bars are ±1 SE.
Average densities of Ecklonia were orders of magnitude higher in the northern region than in the central and southern regions (whole model: F8,2491 = 506.4, r2 = 0.62, p < 0.0001; effects tests: region, F2,2491 = 1279.9, p < 0.0001) (Figure 6a). Marine reserve effect sizes were significantly positive and large (>10–15 individuals/m2) within the northern region and for the shallow and mid depth strata, but not significant in the deep stratum (whole model: F6,241 = 26.2, r2 = 0.40, p < 0.0001; effects tests: reserve, F1,241 = 95.6, p < 0.0001; reserve × depth zone, F2,241 = 13.9, p < 0.0001) (Figure 6b). Marine reserve effects on Ecklonia density were not significant in the central region (whole model: F7,278 = 4.9, r2 = 0.11, p < 0.0001; effects tests: reserve, F1,278 = 1.63, p = 0.20, reserve × depth zone, F2,278 = 2.5, p = 0.08) (Figure 6b). In the southern region, marine reserve effects on Ecklonia density were not significant (F15,1950 = 29.7, r2 = 0.19, p < 0.0001; effects tests: reserve, F1,278 = 0.11, p = 0.73, reserve × depth zone, F2,278 = 4.65, p = 0.009) (Figure 6b). Marine reserve effects on Ecklonia density were significantly large (>10 individuals/m2) in the northern region, and relatively small if significant in the other two regions (F8,2491 = 506.4, r2 = 0.62, p < 0.0001; effects tests: region × reserve, F2,2491 = 274.6, p < 0.0001) (Figure 6c).
Enlarge this image.FIGURE 6. (a) Ecklonia radiata density (number per square meter) for each of the three regions, (b) marine reserve (MR) effect sizes for Ecklonia radiata density by depth zone, an asterisk (*) indicates significant effects, and (c) density of Ecklonia radiata in marine reserves and open areas (OPEN) in each of the three regions. Levels that are not connected by the same letter are significantly different based on Tukey's post hoc tests. Error bars are ±1 SE.
The results of our analysis of all possible combinations of environmental and Evechinus density effects on Ecklonia density yielded three most likely potential effects on which we focused for the AICc analysis: (1) TSS (in grams per cubic meter), (2) regional density of Evechinus chloroticus (number per square meter), and (3) marine reserve effects on site-specific density of Evechinus chloroticus (number per square meter). The TSS covaried among regions with both black disk HVIS distance, the most robust satellite-derived measure of water clarity, and BBP, a measure of the total amount of particulate material in near-surface water. Accordingly, we reduced the model to the single environmental parameter, TSS. The AICc analysis demonstrated that the statistical explanatory power of regional differences in (1) TSS (in grams per cubic meter), (2) average regional density of Evechinus chloroticus (number per square meter), and (3) marine reserve effects on the density of Evechinus chloroticus (number per square meter) on the density of Ecklonia radiata was different among depth zones (0–5, 6–10, and 11–15 m) when we considered patterns across all 63 marine reserve and reference sites. In the shallow depth stratum (0–5 m), marine reserve effects on Evechinus density had the highest factor weighting (1.0) while both TSS and the regional density of Evechinus had factor weightings above 0.9. In the mid depth stratum (6–10 m), TSS had the highest factor weighting (1.0), while marine reserve effects on Evechinus density and the regional density of Evechinus both had factor weightings above 0.9. In the deepest depth stratum (11–15 m), TSS had the highest factor weighting (1.0), while the regional density of Evechinus had a factor weighting of 0.95 and marine reserve effects on Evechinus density had a lower factor weighting of 0.21 (Table 2).
TABLE 2 Results of corrected Akaike information criterion (AICc) analysis by depth zone (a) 0–5 m, (b) 6–10 m, and (c) 11–15 m for the statistical effects of (1) total suspended solids (TSS, in grams per cubic meter), (2) average regional density of
| Model | K | RSS | r2 | AICc | Δi | wi | Factor weight |
| (a) | |||||||
| MREve × EVER × TSS | 5 | 2772.21 | 0.51 | 249.46 | 0 | 0.977 | |
| MREve × TSS | 4 | 3261.69 | 0.42 | 257.34 | 7.880 | 0.019 | |
| MREve | 3 | 3588.62 | 0.36 | 261.08 | 11.616 | 0.003 | 1.0 |
| MREve ×EVER | 4 | 3578.07 | 0.36 | 263.17 | 13.713 | 0.001 | |
| TSS × EVER | 4 | 4051.81 | 0.28 | 271.00 | 21.546 | 2.05E-05 | |
| TSS | 3 | 4576.61 | 0.19 | 276.39 | 26.936 | 1.38E-06 | 0.99 |
| EVER | 3 | 4822.74 | 0.14 | 279.69 | 30.237 | 2.65E-07 | 0.97 |
| (b) | |||||||
| MREve × EVER × TSS | 5 | 2740.56 | 0.52 | 234.77 | 0 | 0.910 | |
| TSS × EVER | 4 | 3117.50 | 0.45 | 239.85 | 5.075 | 0.072 | |
| MREve × TSS | 4 | 3285.71 | 0.42 | 242.89 | 8.123 | 0.016 | |
| TSS | 3 | 3683.44 | 0.35 | 247.21 | 12.440 | 0.002 | 1.0 |
| MREve | 3 | 3890.23 | 0.31 | 250.38 | 15.608 | 0.0003 | 0.92 |
| MREve × EVER | 4 | 3890.16 | 0.31 | 252.69 | 17.918 | 0.0001 | |
| EVER | 3 | 4256.48 | 0.25 | 255.60 | 20.827 | 2.73E-05 | 0.98 |
| (c) | |||||||
| TSS × EVER | 4 | 1048.38 | 0.28 | 168.98 | 0 | 0.762 | |
| MREve × EVER × TSS | 5 | 1054.41 | 0.51 | 171.72 | 2.743 | 0.193 | |
| TSS | 3 | 1246.56 | 0.19 | 175.99 | 7.013 | 0.023 | 1.0 |
| MREve × TSS | 4 | 1217.32 | 0.42 | 177.05 | 8.068 | 0.013 | |
| MREve | 3 | 1329.90 | 0.36 | 179.49 | 10.507 | 0.004 | 0.21 |
| EVER | 3 | 1350.17 | 0.14 | 180.31 | 11.325 | 0.003 | 0.95 |
| MREve × EVER | 4 | 1321.66 | 0.36 | 181.49 | 12.509 | 0.001 | |
The data and results of this study provide empirical support for strong interactions between the century-long legacy effects wrought from differing histories of land clearance, exploitation of high trophic-level predators, and wide-scale proliferation of sea urchins in three important biogeographic marine regions of New Zealand. Depth-specific spatial associations between sea urchins (Evechinus chloroticus) and the dominant kelp (Ecklonia radiata) at a range of different spatial scales, from quadrat to region both within and outside of no-take marine reserves, provided effective proxies for resolving the consequences of grazer dynamics in these systems and how predation on sea urchins as well as long-term changes in the environmental conditions, increases in turbidity, smothering by fine sediments and shoaling of the compensation point likely mediate interactions between large brown algae and sea urchins. Removal of large predatory fishes, rock lobsters, and truncation of their size distributions can release sea urchin populations from the top–down regulation (Babcock et al., 2010; Estes et al., 2004; Steneck et al., 2004), while the development of abundant sea urchin population networks at the regional scale can enhance local larval supply, increasing the probability of large-scale settlement events and proliferation of adult sea urchins even in local populations with reproductive deficits (Wing, 2009, 2011; Wing et al., 2003). Interactions between proxies for these scale-dependent processes were strongly influenced regionally by widespread degradation of the environmental conditions wrought by land-based stressors.
Degradation of coastal water quality from extensive inputs of fine sediments has resulted in the contraction of multi-layered kelp forest habitats as a direct result of the legacy of land conversion at the regional scale (Schiel, 2013; Urlich & Handley, 2020). In addition, nutrient pollution associated with the increases in the application of synthetic nitrogen fertilizers in the 1960s “green revolution” has resulted in a shift in nutrient budgets within kelp forests and other coastal systems (Vitousek et al., 1997). For example, Sabadel et al. (2020) used isotopic analysis of δ15N in phenylalanine, a “source” amino acid, whose isotopic value is reflective of those in the primary producers at the base of the food web, in samples of the herbivorous fish Odax pullus (Odacidae, weed whitings) collected in the 1950s to present day, to resolve large shifts in the coastal nitrogen budget coincident with wide-scale synthetic fertilizer application in New Zealand. Similarly, the development of high-intensity salmon farming near rocky reef habitats in the Marlborough Sounds has resulted in direct inputs of nutrients to kelp forest food webs (McMullin & Wing, 2021). The legacy of eutrophication and shifts in the nitrogen cycle brought by these and other anthropogenic inputs can directly influence the structure and composition of kelp forest systems (Dayton et al., 1992), as well as the life cycle of organisms with planktotrophic larvae such as echinoderms.
Our analysis of the southern region, comprising sites throughout Fiordland, provided an effective reference for relatively small effect sizes of the legacies of land-based stressors, trophic downgrading, and regional proliferation of sea urchins. The Fiordland region is covered in mature stands of native podocarp and Nothofagus beech forest, and except for the influence of a large hydroelectric power station in one Sound (Jack et al., 2009; Tallis et al., 2004), has unmodified catchments (Wing & Jack, 2014). Rock lobster stocks have been depleted but are recovering after the implementation of the Fiordland Marine Management Act in 2005, and the density of potential predators of Evechinus is high compared with the other regions considered in this study (Udy, Wing, Jowett, et al., 2019). Ecklonia was found at relatively high densities in the deepest depth strata of subtidal surveys (Figure 4b), which is consistent with deep penetration of PAR in a clear water column at the fjord entrances and low levels of smothering by fine sediments (Wing et al., 2007). Marine reserve effect on and densities of Evechinus within the region are relatively small (Figure 5), and the regional population consists of a series of reproductive sources and sinks (Wing, 2009, 2003). Ecklonia occurs throughout Fiordland but nevertheless densities are typically low but distributed to deep zones (>20 m) in topographically shaded, low wave-energy sites (Miller et al., 2000; Wing et al., 2007). Sites along the outer coast are characterized by clear, subtropical water supplied by the Fiordland current (Chandler et al., 2021) and moderate-to-high oceanic wave exposure (Wing et al., 2007).
The patterns in Fiordland are in sharp contrast to those observed in the central region. The catchments surrounding Marlborough Sounds and Tasman Bay have undergone extensive land-based disturbance with the conversion of coastal lands to farming and forestry particularly over the last century (Udy, Wing, O'Connell-Milne, et al., 2019; Urlich & Handley, 2020). Combined with marine heat waves, extensive fine sedimentation has driven a wide-scale decline in large brown algae, most obviously surface canopies of Macrocystis pyrifera, from all but the oceanic-influenced entrances of the Sounds over at least the last 50 years (Handley, 2016; Hay, 1990; Tait et al., 2021). In addition to the diffuse effects of land-based inputs, overharvest, decline, and truncation of the size distributions of sea urchin predators such as red rock lobsters (J. edwardsii), blue cod (P. colias), and snapper (P. auratus) (Davidson et al., 2014; Kolodzey & Wing, 2022) have coincided with the regional proliferation of the sea urchin population. Consequentially, there is a large regional population size of Evechinus, relatively small marine reserve effect sizes, and a near absence of Ecklonia beyond 11-m depth on rocky reefs. Combined with the high TSS and poor underwater visibility, these patterns are consistent with a strong influence of land-based stressors in the region, shoaling of the compensation point for kelp photosynthesis and smothering of new recruits by fine sediments, and regional expansion of the Evechinus metapopulation. Interactions between the two primary legacy effects in the region have resulted in a degraded system at the regional level, which shows only moderately positive responses in kelp density in the absence of fishing pressure in the marine reserves.
The observed patterns in the central region are likely influenced by the increased susceptibility of physiologically stressed kelp to overgrazing (Udy, Wing, Jowett, et al., 2019). Theory predicts that grazing distributed along gradients in primary production can result in sharp thresholds for persistence as grazing pressure exceeds vegetative regeneration (Oksanen et al., 1981). The effect has been observed in kelp forest systems where increases in the supply of detrital organic matter from kelps have suppressed deforestation from grazing sea urchins (Rennick et al., 2022). Physical oceanographic conditions in the region are dominated by tidal flows that result in high retention of neritic water and larvae in the inner Sounds and in persistent retention of sediment loads in eddies within Tasman Bay (Heath, 1976; Stevens et al., 2021).
Observations of strong marine reserve effects on Ecklonia abundance despite an abundant regional population of Evechinus indicate that trophic processes are likely dominant within the northern region around Leigh. High densities of Ecklonia observed in the deepest depth strata of the survey in the absence of Evechinus are consistent with the absence of light limitation or smothering by fine sediments on the deep reef. These observations are consistent with a long history of fisheries exploitation in the region and the proximity of the sites to a major population center (Auckland, population 1.6 million), which has resulted in the loss of some of the large predators on sea urchins in kelp forests, particularly large individuals of New Zealand snapper (P. auratus), the red rock lobster (J. edwardsii), and packhorse lobster (Sagmariasus verreauxi). Relatively clear subtropical waters are replenished at the sites by the East Auckland Current (Stanton et al., 1997), which combined with oceanic wave exposure creates much better growth conditions at the sites surveyed than in the other two regions for dense stands of Ecklonia.
The patterns we present are field-based and observational and likely reflect contrasting long-term ecological conditions among the three regions. The approach was correlative, and we employed strong inference supported by myriad studies rather than direct experimental tests to infer likely processes from patterns that could have been well-explained by a variety of drivers. Nevertheless, we used empirical data to provide resolution of the dynamics acting in the system relative to two strong categories of regional legacy effects, and we present results as consistent with a process rather than a falsified test of that process, as support for further experimental work. Although Evechinus and Ecklonia occur in all regions, they are components of biogeographically different subtidal communities and are subjected to the different environmental conditions among regions. The histories of land conversion, geomorphology, local oceanographic conditions, turbidity, and fishing pressure in the northern, central, and southern regions are dissimilar, and we have focused on relative differences in patterns of distribution among regions rather than primarily on absolute densities. Within the limitations of a correlative, observational study such as this, the observed patterns are useful for considering the relative influences of the long-term legacies wrought by the loss of large predators, clearance of native forests as well as both regional and local patterns in the drivers of the proliferation of sea urchins on the abundance patterns of Ecklonia-dominated kelp forests in New Zealand.
Evechinus and Ecklonia have potentially strong ecological interactions that are strongly influenced by the environmental conditions and can result in state changes on rocky reefs. The likelihood of state changes can be modified by changes in the physical environment wrought from inputs of fine sediments that can smother and reduce light penetration to subtidal reefs, and by regional-scale release of Evechinus populations from regulation by predators, including several actively exploited species. Observations of the signatures of strong legacy effects associated with land clearance, trophic structure changes, and their interaction provide a useful platform to inform regional approaches to ecosystem-based management and application of appropriate measures to mitigate or prioritize conservation of intact kelp forest ecosystems in New Zealand.
ACKNOWLEDGMENTSWe gratefully acknowledge the work of our subtidal research diving teams, particularly, Jacquetta Udy, Rebecca McMullin, Leo Durante, Clara Schlieman, Charlotte Borra, Sara Rutger, Lucy Wing, Sorrel O'Connell-Milne, Kim Clark, and the Masters of research vessels Bill Dickson, Bob Walker, and Pete Young. Support for the research was provided by the Royal Society of New Zealand's Marsden Fund, the Department of Conservation, the National Science Challenge: Sustainable Seas under projects 4.1.1 and 1.1, and Ministry of Business, Innovation and Employment contract UOCX1704.
CONFLICT OF INTERESTThe authors declare no conflict of interest.
DATA AVAILABILITY STATEMENTData (Wing, Durante, et al., 2022; Wing, Shears, et al., 2022) are available from Dryad:
© 2022. This work is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Details
; Shears, Nicolas T 2 ; Tait, Leigh W 3 ; Schiel, David R 4 1 Department of Marine Science, University of Otago, Dunedin, New Zealand
2 Department of Statistics, University of Auckland, Auckland, New Zealand
3 National Institute for Water and Atmospheric Research, Christchurch, New Zealand
4 Department of Biological Science, Canterbury University, Christchurch, New Zealand