Science of the Total Environment 896 (2023) 164829

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Effects of connectivity on carbon and nitrogen stocks in mangrove and

seagrass ecosystems
⁎
Daniel Saavedra-Hortua a,b, , Ivan Nagelkerken c, Lina M. Estupinan-Suarez d, Lucy Gwen Gillis e
a
Leibniz Centre for Tropical Marine Research (ZMT) Fahrenheitstraße 6, 28359 Bremen GmbH, Germany
b
ecosecurities, Rue de la Faïencerie 2, 1227, Carouge, Switzerland
c
Southern Seas Ecology Laboratories, School of Biological Sciences, The University of Adelaide, DX 650 418, Adelaide, SA, 5005, Australia
d
Max Planck Institute for Biogeochemistry, Hans-Knöll-Str. 10, 07745 Jena, Germany
e
Hydro-Environmental Services, IMDC, 2000, Antwerpen, Belgium

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• Connectivity of mangroves-seagrass in-

creases carbon sequestration in mangrove
patches.
• Internal attributes enhance seagrass car-
bon sequestration.
• Connectivity needs to be considered for
enhancement of ecosystems services.

A R T I C L E I N F O A B S T R A C T

Editor: Paulo Pereira Seascape connectivity increases carbon and nitrogen exchange across coastal ecosystems through flow of particulate
organic matter (POM). However, there are still critical gaps in knowledge about the drivers that mediate these pro-
Keywords: cesses, especially at regional seascape scales. The aim of this study was to associate three seascape-level drivers
Seascape ecology which could influence carbon and nitrogen stocks in intertidal coastal seascape: connectivity between ecosystems, eco-
Nutrient exchange
system surface area, and standing vegetation biomass of ecosystems. Firstly, we compared whether connected man-
Blue carbon dynamics
Nitrogen retention
grove and seagrass ecosystems contain larger carbon and nitrogen storage than isolated mangrove and seagrass
Connectivity ecosystems. Secondly, we compared autochthonous and allochthonous POM in mangrove patches and seagrass
Coastal vegetated ecosystems beds, simultaneously estimating the area and biomass relative contribution to POM of the different coastal vegetated
ecosystem. Connected vs isolated mangrove and seagrass ecosystems were studied at six locations in a temperate sea-
scape, and their carbon and nitrogen content in the standing vegetation biomass and sediments were measured. POM
contributions of these and surrounding ecosystems were determined using stable isotopic tracers. In connected
mangrove-seagrass seascapes, mangroves occupied 3 % of total coastal ecosystem surface area, however, their stand-
ing biomass carbon content and nitrogen per unit area was 9–12 times higher than seagrasses and twice as high as
macroalgal beds (both in connected and isolated seascapes). Additionally in connected mangrove-seagrass seascapes,
the largest contributors to POM were mangroves (10–50 %) and macroalgal beds (20–50 %). In isolated seagrasses,
seagrass (37–77 %) and macroalgal thalli (9–43 %) contributed the most, whilst in the isolated mangrove, salt marshes
were the main contributor (17–47 %). Seagrass connectivity enhances mangrove carbon sequestration per unit area,
whilst internal attributes enhance seagrass carbon sequestration. Mangroves and macroalgal beds are potential critical

⁎ Corresponding author.
E-mail addresses: daniel.hortua@leibniz-zmt.de dashortua@gmail.com (D. Saavedra-Hortua).

http://dx.doi.org/10.1016/j.scitotenv.2023.164829
Received 4 September 2022; Received in revised form 31 May 2023; Accepted 10 June 2023
Available online 14 June 2023
0048-9697/© 2023 Published by Elsevier B.V.
D. Saavedra-Hortua et al.
contributors of nitrogen and carbon to other ecosystems. Considering all ecosystems as a continuing system with
seascape-level connectivity will support management and improve knowledge of critical ecosystem services.
1. Introduction
Coastal vegetated ecosystems (CVE), such as mangroves, salt marshes,
seagrass and macroalgal beds, are key marine ecosystems for carbon se-
questration and climate change adaptation (Duarte et al., 2013a, Saderne
et al., 2019). Two main characteristics make CVE important carbon sinks,
which could help mitigate rapidly increasing anthropogenic CO2 levels in
the atmosphere. Firstly, average plant communities in CVE have a higher
rate of primary production than respiration rate, accumulating organic car-
bon in the biomass (Duarte et al., 2005). Secondly, enhanced carbon se-
questration occurs in the sediments of mangroves, salt marshes and
seagrass beds as a result of reduced water flow through the plant roots
and shoots, which increases particle trapping capacity (Kennedy et al.,
2010; Duarte et al., 2013a; Alongi, 2014). For example, global carbon
burial rates range between 5 and 87 Tg C yr−1 in salt marshes, 31–34 Tg
C yr−1 in mangroves, 48–112 Tg C yr−1 in seagrass beds and 61–268 Tg
C yr−1 in macroalgal beds, compared to 49.3–78.5 Tg C yr−1 in terrestrial
forests (McLeod et al., 2011; Krause-Jensen and Duarte, 2016). Because
sediment carbon sequestration originates from allochthonous and autoch-
thonous sources (Kennedy et al., 2010; Alongi, 2014; Gillis et al., 2014a),
a seascape perspective is crucial for understanding the importance of the
different CVE and their role as carbon donors and recipients.
Nutrient retention is one important ecosystem service provided by
wetlands and coastal wetlands (Mitsch et al., 2014; Pei et al., 2020). Plants
and soils can retain significant amounts of nitrogen, therefore preventing ni-
trogen for being released into the surrounding environments (de Vries and
Bardgett, 2016). Nitrogen retention is an important coastal ecosystem service
that helps to diminish the negative effects of high nitrogen concentrations in
the coastal waters and in the ocean (Saunders and Kalff, 2001). Coastal
vegetated ecosystems play an important role in retaining nitrogen within
the ecosystem (Reis et al., 2017; Kindeberg et al., 2018; Santos et al.,
2019). Nitrogen retained in standing biomass of seagrasses and macroalgae
provides a short-term (up to 2 years) buffer to nutrient loading effects in
the water column (McGlathery et al., 2007). Mangrove trees are capable of
retaining more nitrogen and for longer time periods than seagrasses,
macroalgae, and salt marshes due to a bigger and more robust plant
structure. Mangroves appear to have higher sediment nitrogen contents
(12.2–15.4 Mg N ha−1; Reis et al., 2017) compared to terrestrial ecosystems
(0.5–2.5 Mg N ha−1; Xu et al., 2019) and seagrass beds (0.2–4.5 Mg N ha−1;
McGlathery et al., 2007; Kindeberg et al., 2018). Macroalgal beds have one of
the highest primary production rates in the coastal zone (420 NPP g C m−2
y−1) (Krause-Jensen and Duarte, 2016), increasing the carbon and nitrogen
storage capacity within the standing biomass. However, because macroalgal
beds often grow on hard, rocky substrate, they do not develop significant sed-
iment deposits compared to other CVE (Duarte et al., 2013a; Hill et al.,
2015). The capacity of CVE in retaining organic carbon in their sediments
may also support retention of nutrients such as nitrogen and phosphorous,
as these nutrients and carbon follow the Redfield ratio (nearly constant
ratio of C:N:P 106:16:1 across ocean habitats) and therefore are stoichiomet-
rically linked (Saderne et al., 2020).
Seascape connectivity in terms of the interchange of energy (i.e., waves)
and materials (e.g., sediments, particulate/dissolved nutrients) influences
processes such as ecosystem productivity, and trapping and storage of
carbon and nutrients inside CVE (Gillis et al., 2014a; Huxham et al., 2018).
Except for macroalgal beds, CVE mainly inhabit sedimentary landscapes
(Duarte et al., 2013a). Sedimentary CVE receive and accumulate sediments
with a high amount of minerals, carbon and nutrients through tidal flows
(Woodroffe et al., 2016; Rogers et al., 2019). Physiological process related
with primary production could influence carbon and nitrogen accumulation
within the ecosystems, for example mangroves and salt marshes have higher
2
Science of the Total Environment 896 (2023) 164829
net primary production (394–1000 g C m−2 yr−1) rate compared with
seagrass and macroalgae beds (91–552 g C m−2 yr−1), however, burial and
export rates vary extremely across the globe (Duarte et al., 2005). Organic
sediments in CVE can be of autochthonous origin – from within the ecosystem
due to retention of detrital material – or allochthonous origin – via inputs of
particulate organic matter (POM) from connected ecosystems (Hyndes et al.,
2014). Through reducing wave action, CVE are efficient in sediment deposi-
tion, allowing for carbon and nitrogen-rich POM accumulation (Saderne
et al., 2019; de los Santos et al., 2022; Santos et al., 2019). Mangrove forests
and salt marshes often outwell nutrients to connected seagrass beds (Kennedy
et al., 2010; Hyndes et al., 2014), but some mangrove forests also experience
inwelling (Walton et al., 2014). Ecosystem attributes such as area covered by
vegetation are important in modulating the capacity for carbon exchange and
storage in CVE (Adame and Lovelock, 2011; Serrano et al., 2016; Rogers
et al., 2019) and nitrogen removal or storage (Valiela and Cole, 2002). Re-
cently, it has been demonstrated that mangroves with a high carbon stock
could have low carbon sequestration rates and vice versa (Jennerjahn,
2020). This implies that besides carbon stock it is crucial to evaluate processes
at the landscape level such as exchange of POM and sediment accumulation
of allochthonous organic matter. The majority of research on connectivity
in the tropical seascape has concentrated on exchange of material, energy,
and organisms, between mangrove forests and seagrass beds (Nagelkerken,
2009; Chen et al., 2017; Huxham et al., 2018).
Recently, studies have highlighted the importance of the contribution
from other CVE within the seascape, such as macroalgal beds and salt
marshes (Santos et al., 2019; de los Santos et al., 2022). Other factors such
as ecosystem surface area and standing biomass of CVE could also strongly
influence the quantity of POM that an ecosystem produces and potentially in-
terchanges with other CVE across the coastal seascape. Surface area has been
explored only by a few studies (Gillis et al., 2014b, Saavedra-Hortua et al.,
2019; Güllstrom et al., 2017; Phang et al., 2015). These studies have founded
that ecosystems such as mangrove forests and seagrass beds could contribute
significantly to the POM even if they occupy a relatively small area compared
with other ecosystems within the seascape. Nevertheless, to the best of our
knowledge biomass standing influence on the ecosystem's contribution to
the POM has rarely been investigated. This information is essential to under-
stand the full capacity of coastal ecosystem ability to store carbon and nitro-
gen, and the ecosystems services they provide on a regional seascape scale.
Understanding the drivers of nitrogen and carbon storage in coastal veg-
etated ecosystems is critical to manage associated ecosystem services for car-
bon mitigation and nutrient retention. Coastal vegetated ecosystems are part
of an interconnected seascape of matter exchange, but we still lack insight
into how this seascape connectivity alters the magnitude and direction of car-
bon and nitrogen export, deposition, and storage compared to ecosystems
that exist in isolation (Huxham et al., 2018; Santos et al., 2019). There are
three main gaps we have identified which require further investigations: con-
nectivity between ecosystems, surface area of ecosystems and standing bio-
mass of ecosystems. The aim of this study was to clarify how these three
factors influence nitrogen and carbon storage of coastal vegetated ecosys-
tems. To investigate carbon and nitrogen stocks across connected seascapes,
we firstly tested whether connected mangrove and seagrass ecosystems con-
tain larger carbon and nitrogen stocks compared to those isolated from other
ecosystems. Secondly, to understand how the biomass and area of the differ-
ent CVE influence the POM input of each CVE within the seascape, we calcu-
lated carbon and nitrogen stocks across three types of mangrove patch and
seagrass bed combinations in South Australia: i) connected mangrove
patches and seagrass beds (connected to each other within 3 km), ii) isolated
seagrass beds (separated from mangrove forests at least 3 km apart), iii) and
isolated mangrove patches (separated from seagrass beds at least 3 km
apart). To test the inputs from surrounding ecosystems to the POM in
D. Saavedra-Hortua et al. Science of the Total Environment 896 (2023) 164829

mangrove patches and seagrass beds, we measured the contribution of al-
lochthonous vs. autochthonous sources of particulate organic matter across
the four CVE (mangrove patches, seagrass beds, macroalgal beds and salt
marshes) and related this to their surface area and standing biomass. By
studying the differences in carbon and nitrogen stock as well as POM ex-
change across CVE, together with their surface area and standing biomass,
we can clearly understand more the critical role and impact of individual
CVE on ecosystems services of carbon sequestration and nitrogen retention,
from a seascape perspective.
2. Materials and methods
2.1. Study site
The study site was located in Gulf St. Vincent and Murray River dis-
charge region which have predominantly semi-arid and arid climate
(Edyvane, 1999), Gulf St Vincent coastal and intertidal vegetation comprise
mainly seagrass beds, mangrove forest and salt marshes (Edyvane, 1996,
1999). Additionally, In Gulf St. Vincent there are oceanographic processes
dominated by a high tidal range, therefore forming extensive intertidal
areas. Shallow depths, with very low terrigenous input and high evapora-
tion (Richardson et al., 2005). According to the analysis of land cover
areas of the different (See description below in section 2.3) the catchment
areas of our locations were 4723 km2 along the coast of Gulf St Vincent,
872 km2 in Rushy Island and 366 km2 in Victor Harbor. The isolated man-
grove patch was dominated by salt marshes (99.6 % of the total ecosystem
area), whilst in the isolated seagrass bed, macroalgal beds also dominated
(90.8 % of the total ecosystem area).
outside the Gulf) across the South Australian coast (Fig. 1). We chose loca-
tions which represented different combinations of isolated vs connected
mangrove patches and seagrass beds. We only sampled in the mangrove
patches and seagrass beds as this information was initially used for quanti-
fying the carbon and nitrogen stocks in connected mangrove patches and
seagrass beds. At each location, three parallel transects were established
during low tide, covering a representative area of the ecosystem. Transects
were separated between 50 and 400 m. Transects had a length between 400
and 1200 m covering the majority of the CVE area, starting at the ocean side
of the tidal flat or subtidal seagrass bed zones, stretching landwards
through the intertidal seagrass bed and/or mangrove patch and continuing
up to the salt marsh or terrestrial zone, depending on the location (Fig. 1).
Each transect had between five and six transect-points separated by 70 to
200 m. Mangroves connected to seagrass beds were sampled at low tide
close to spring tides times to extend the sampling area oceanward, whilst
mangrove patches were sampled during low neap tides. Field sampling at
each location was completed within two to four days. All samples were
taken during low tide, except for suspended particulate matter that were
taken during high tide, when the tide was flowing out.
We decided to include macroalgal beds and salt marshes in our ecosys-
tems surface area calculations (Fig. 1), because of the wide-spread distribu-
tion at the seascape level of these ecosystems and their potential
importance as a carbon or nitrogen source (Krause-jensen et al., 2018;
Ortega et al., 2019; Santos et al., 2019). Therefore, we included them in
our analysis as a source of nitrogen and carbon in the seascape even though
we did not directly sample these systems.
2.3. Coastal vegetated ecosystems mapping

2.2. Sample design Coordinates of ecosystem boundaries in Gulf St. Vincent were latitude
34°3 to 35°24′ S, and longitude 137°40′ to 138°40′ E, whilst Victor Harbor
Sampling and measurements were performed during the dry season coordinates were latitude 35°27′ to 35°40′ S and longitude 138°31′ to
(April–May 2017) at six locations (four inside Gulf St. Vincent and two 138°38′ E and Rushy Island 35°28′ to 35°32′ S and longitude 138°46′ to

Fig. 1. Blue circles ( ) indicate connected mangrove patches and seagrass beds (Port Clinton, Port Wakefield and Torrens Island), the yellow circles ( ) indicate seagrass beds
with no connected mangrove patches (Victor Harbor and Stansbury), and the light green circle ( ) indicates a mangrove patch with no connected seagrass bed (Rushy Island).
The light grey area indicates the catchment zone of the Gulf St. Vincent locations, the white area that of Victor Harbor, and the dark grey area that of Rushy Island. Sampling
design in each location is showed in the small boxes on the right side. In rushy Island mangrove patches were not really disperse and no detected by land cover reported by
Willoughby et al., 2018, however small patches pictures were in the location (Supplementary fig. 2 Panel C).

3
D. Saavedra-Hortua et al.
138°52′ E. Mangrove patches and salt marshes were selected from the
Department for Environment and Water of South Australia (DEW) land
cover map (Willoughby et al., 2018) coast to assess exclusively vegetation
that influences the coastal measurements within a 3 km limit landwards
from the coastline. The land cover map was based on a predictive model
of native vegetation using Landsat imagery, local calibration data,
and other supplemented datasets, with a minimum scale of 1:50.000
(Willoughby et al., 2018). Land cover maps and watersheds (catchment
area) were acquired from the nature maps web server of (DEW, Dec.
2019). The limits of the cover area of CVE (mangrove patches, seagrass
beds, macroalgal beds and salt marshes) taking for the calculation of eco-
systems surface area and standing biomass in St. Vincent Gulf, Rushy Island
and Victor harbor were determined based on the Bathymetry lines showed
in http://spatialwebapps.environment.sa.gov.au/naturemaps/, Dec. 2019.
The South Australian Marine Benthic Habitats map was provided under li-
cense number 2696 by the same agency. http://spatialwebapps.
environment.sa.gov.au/naturemaps/, Dec. 2019. The limits of the cover
area of CVE (mangrove patches, seagrass beds, macroalgal beds and salt
marshes) which were taken for the calculation of the ecosystems surface
area and standing biomass in St. Vincent Gulf, Rushy Island and Victor har-
bor were determined based on the Bathymetry lines showed in http://
spatialwebapps.environment.sa.gov.au/naturemaps/, Dec. 2019. The
South Australian Marine Benthic Habitats map was provided under license
number 2696 by the same agency.
Cover area of seagrass and macroalgal beds were obtained from the
Marine Benthic Habitats map (DEW, 2019). Data obtained from DEW 2017
were digitized from discernible underwater features of ortho-rectified aerial
photographs (2005–2007), field observations and underwater video footage
incorporated. Data was updated in 2014, and its scale ranges from 1:5.000
to 1:10.000. For the seagrasses and macroalgal ecosystems, we summed all
four vegetation cover classes: continuous dense, continuous medium, patchy
dense and patchy medium, as provided in the Marine Benthic Habitats map.
Watershed selection was based on the sampling point locations to account
for the area of terrestrial influence. Data layers were projected to GDA94 /
SA Lambert - EPSG:3107 and converted from raster to shapefiles when neces-
sary. The area of all vegetation classes specified above was calculated in km2.
The geographical analysis was run in ArcGIS version 10.5. Description of data
used for mapping is showed in Supplementary Table 1.
2.4. Standing biomass calculation
To quantify the relative dimensionless contribution indices for total above-
ground biomass of each system an initial first step was to calculate the stand-
ing biomass of our mangrove patches and seagrass sites. In the mangrove
patches, at each transect point a plot of 100 m2 was established and the diam-
eter at breast height (DBH) approximately at 1.3 m of each mangrove tree was
measured. This was used to estimate the standing biomass (Howard et al.,
2014), hereafter referred to as ‘biomass’. The only species found in the patches
was Avicennia marina. Aboveground and belowground biomass (AGB and
BGB, respectively) was estimated from the DBH, using region-specific allome-
tric equations reported in Bulmer et al. (2016a). This was the allometric equa-
tion developed for the same mangrove species in New Zealand. In Port
Wakefield some trees closer to the terrestrial zone had heights lower than
1.3 m; therefore, allometric equations reported by Bulmer et al. (2016b)
were used to calculate biomass at this location.
In the seagrass beds, at each transect point a core (15 cm diameter, total
area of 176 cm2 and 20 cm depth) was taken to sample seagrass biomass.
Seagrass plants sampled in the core were separated into AGB (shoots) and
BGB (roots and rhizomes), cleaned from attached sediment and epiphytes
with seawater and further rinsed with distilled water in the laboratory.
Samples were dried for approximately 72 h at 60 °C until constant weight,
this weight was used to calculate the seagrass AGB and BGB for each
transect point. In St. Vincent gulf the seagrass species found were Heterozostera
tasmanica, Zostera muelleri, Zostera mucronata, whilst in Victor harbor
Amphibolis antartica, Amphibolis griffithii and Posidonia sinuosa were more
common.
4
Science of the Total Environment 896 (2023) 164829
To quantify the relative dimensionless contribution indices for macroalgal
and salt marsh plants biomass of each system an initial first step was to calcu-
late the standing biomass of the sites. Biomass of salt marsh plants and
macroalgal thalli were not directly measured, therefore the values were
taken from studies in a close location with similar characteristics (Hill et al.,
2015 and references therein; Serrano et al., 2019, and references therein).
The macroalgal species that we found in seagrass beds and mangrove patches
were Hormosira banksia and Ulva sp., which also have been reported in the
southern part of Australia (Phillips, 2001). Salt marshes species founded and
collected in Rushy island were Juncus kraussi and Juncus sp., whilst in Gulf
St. Vincent Sporobolus virginicus and Sarcocornia quinqueflora have been re-
ported to be very abundant (Connolly et al., 2005).
2.5. Sediment
2.5.1. Core sampling
To quantify carbon and nitrogen content in the sediment, we took a sed-
iment core of 15 cm depth and 7 cm diameter. These cores were taken dur-
ing low tide at each transect point. Each core was divided into three
subsamples of 0–5, 5–10 and 10–15 cm sediment depth, respectively. Sam-
ples were kept in freezer bags with freezer blocks and transported to the
laboratory within 4 h. Each sediment sample was dried at 60 °C for 72 h
or until constant weight. Sediment density (refer in other studies as soil
bulk density) was calculated by dividing the weight of each sample by the
volume of each sample (192 cm3).


Sediment density ¼ mass of dried soilðgÞ=original volume cm3
Dried samples were transported to the analytical laboratory of the
Leibniz-Centre for Tropical Marine Research (ZMT), Germany, to measure
the percentage of total carbon and nitrogen in the sediment.
2.5.2. Suspended material sampling
To understand sediment fluxes from connected and isolated systems we
analysed suspended sediment, which were taken along transects at each
site. At each point along the transects, two types of suspended sediment
samples were collected, one from the suspended particulate matter in the
water column (SPM), and one from sediment traps. To measure SPM, 1 L
of water was sampled from the upper 40 cm of the water column at each
transect-point during ebb tide. Samples were kept in freezer bags with
freezer blocks and transported to the laboratory within 4 h, where they
were filtered onto pre-combusted (450 °C, 24 h) glass fiber filters (GF/C,
696 grade, 1.2 μm pore diameter). The filters were dried for approximately
72 h at 60 °C until a constant weight.
Sediment traps were made of plastic cylinders of 24 cm length and 6 cm
diameter, with twenty-four 0.5 cm-diameter holes evenly distributed in the
upper 10 cm of the cylinder, and a lid at the bottom collecting the suspended
particulate matter that entered the trap. Sediment traps were attached to steel
rods anchored in the sediment at a height of 5 cm above the substratum. Sed-
iment traps were deployed during low tide and the suspended sediment
trapped samples were collected after 24 h at the following low tide. Samples
were kept in freezer bags with freezer blocks and transported to the laboratory
within 4 h. To separate the water from the sediments and avoid salt contami-
nation, samples were deposited into 200 mL beakers, dried in the oven at
60 °C for 72 h, supernatant water was carefully removed, and an additional
150 mL of distilled water was added to dissolve remaining salt. This process
was repeated until salt particles were not detected visually in the dry sediment
sample. It should be noted that a higher contribution of mangrove trees, salt
marshes and terrestrial plant sources in suspended particulate matter could
have resulted from sampling during ebb tide, whereas sediment traps captured
particulate sediments over two ebb and two flood tides. Dried suspended par-
ticulate matter filters and sediment from the sediment traps were transported
to the analytical laboratory of the Leibniz-Centre for Tropical Marine Research
(ZMT), Germany. Dried suspended particulate matter filters and sediment
from the sediment traps samples were then analysed for isotopic composition
(see section below).
D. Saavedra-Hortua et al.
2.6. Carbon and nitrogen content of sediment and plant material
All sediment sub-samples of each depth inside the core were homoge-
nized, and approximately 15 mg was wrapped in tin caps to measure total
carbon (TC) and total nitrogen (TN) percentage. Both were analysed by
combustion in an elemental analyser (EuroVector EA 3000), using acetani-
lide 4 (71.6 % C) for calibration and a low soil standard (OAS4) as standard
with a precision of 0.06 % of carbon and 0.01 % for TN. We analysed carbon
and nitrogen content in the plant material to calculate the carbon and nitro-
gen stocks in the standing biomass of each CVE. The different plant tissues
at each location were analysed separately for carbon and nitrogen content.
Leaf samples were homogenized but were not acidified, they were also
analysed for organic carbon and TN by combustion in an elemental analyser
(EuroVector EA 3000). Measurements had a precision of 0.36 % for Organic
carbon and 0.05 % for TN for plant material samples. Apple leaf standard
SRM1515 were used as reference material.
2.7. Carbon and nitrogen stocks calculation
The carbon and nitrogen stocks of Gulf St. Vincent, Rushy Island and
Victor Harbor were calculated adding up the sediment and standing
biomass stocks of each CVE separately and multiplying the total carbon or
nitrogen quantity per area for the area occupied of each CVE.
2.7.1. Sediment
Carbon and nitrogen sediment stock calculation of mangrove patches
and seagrass beds was calculated by multiplying the average of % carbon
or nitrogen content of each depth subsample with the sediment density at
that subsample depth and later summing the subsamples values (Howard
et al., 2014). Salt marshes and macroalgae calculation of carbon content
within the sediments were done based on data of % carbon and biomass
per unit area reported in Serrano et al. (2019) and references therein for
salt marshes and in Hill et al. (2015) and references therein for macroalgae.
2.7.2. Standing biomass
Carbon and nitrogen content calculation of mangrove patches was
calculated by multiplying the average of % content of species with the
biomass at each point. For seagrass plants carbon content was calculated
multiplying the biomass value of the seagrass core with the average of %
carbon content of species (Howard et al., 2014). Salt marshes and
macroalgae calculation of carbon content within the biomass were done
based on data of % carbon and biomass per unit area reported in Serrano
et al. (2019) and references therein for salt marshes and in Copertino
et al. (2005) and Hill et al. (2015) and references therein for macroalgae.
2.8. Carbon and nitrogen stable isotope signatures
To determine the exchanges between ecosystems, stable isotopes were
used as a chemical tracer. For isotopic signatures of carbon and nitrogen,
six vegetation sources of particulate organic matter (POM) were considered
in this study: oceanic (phytoplankton, zooplankton, and associated
particles), mangrove trees, seagrass plants, terrestrial plants, macroalgal
thalli and salt marsh plants. At each location, fresh plant leaves (4–5) and
macroalgal thalli of the most abundant species were taken and placed in
separate sample bags to transport to the laboratory. The most abundant
species are as follows: mangrove trees (Avicennia marina), seagrasses
(Posidonia sinousa, Zostera mucronata, Heterozostera tasmanica), terrestrial
plants (Maleluca sp., Danthonia sp., Olearia axillaris), macroalgae thalli
(Hormosira banksii, Ulva sp.) and salt marshes (Juncus kraussi, Juncus sp.).
The leaves and thalli were rinsed with distilled water and dried at 60 °C
for 72 h to constant weight. Some sources were missing δ13C and δ15N
data, and these were taken from the literature. For example, oceanic
sources from studies done in Victoria and South Australia (Davenport and
Bax, 2002) and salt marshes isotopic signal of pooled samples of Sporobolus
virginicus and Sarcocornia quinqueflora in Gulf St. Vincent (Connolly et al.,
2005). Additionally, in Torrens Island seagrass beds, due to weather
5
Science of the Total Environment 896 (2023) 164829
conditions only one sample of macroalgal thalli could be collected, and
therefore δ15N signatures were used from Connolly et al. (2013), and
mean δ13C values from Gulf St. Vincent (Port Wakefield and Port Clinton)
were used.
The different plant tissues (sources) at each location, and the two types
of particulate organic matter samples (suspended particulate matter and
sediment trap) were analysed separately for isotope ratios of carbon and ni-
trogen. Isotope ratios were expressed in the delta notation (δ13C, δ15N) rel-
ative to Vienna PDB and atmospheric nitrogen. Ratios of δ13C and δ15N
were determined using a gas isotope ratio mass spectrometer (Thermo
Finnigan Delta Plus) after high temperature combustion in an elemental
analyser (Flash 1112 EA). Analytical precision was ±0.10 ‰ for δ15N
and 0.13 ‰ for δ13C, as estimated from an international standard (Peptone)
analysed together with the samples.
2.9. Data analysis
To investigate carbon and nitrogen stocks across connected mangrove
patches and seagrass beds versus isolated ecosystems in the Gulf St. Vincent
region, generalized linear models (GLMs) were used with the function glmer
using the formula (x ~ Connectivity type + Ecosystem +1│Location) in R
version 3.0.2 (R Core Team, 2013), using the ´lm4´ package (Bates et al.,
2015), where x was the response variable (e.g., carbon percentage),
connectivity type was a factor with 2 levels (connected and isolated),
ecosystem was a factor with 2 levels (mangrove patch, seagrass bed) and
location was a random factor with 6 levels. Differences were considered
statistically significant at the P < 0.05 level. Percentage data was log-
transformed using link = log inside the glmer function, continuous data
were not transformed and the link = identity was used, percentage and
continuous data were analysed using distribution family = gamma (Zuur
et al., 2009).
To estimate the inputs from surrounding coastal ecosystems, a Bayesian
mixing model MixSIAR using stable δ13C and δ15N isotope signatures of the
six vegetation sources was used. This approach allows for the use of few
more sources than tracers (Stock et al., 2018). The mixing model was
used for estimating the percent contribution of these sources to the POM
(suspended particulate matter and sediment traps) sampled at each transect
point for each location. Two relative dimensionless contribution indices
(habitat surface area and biomass) were calculated. Indices were calculated
for total area and biomass, by dividing the contribution of each source as
provided by the mixing model (% Contribution) by the relative percentage
surface area (% Area) or the percent of biomass (% Biomass) of the respec-
tive CVE (mangrove patch, salt marsh, seagrass bed, or macroalgal bed, re-
spectively) as a function of the total seascape surface area at Gulf St. Vincent
for the locations placed there, and in an area of approximately of 200 km2
for the isolated CVE in Rushy Island and Victor Harbor. Due to the hydrody-
namic processes within the Gulf St. Vincent, for the calculation of the two
relative dimensionless contribution indices we used the whole area or bio-
mass inside the Gulf or in a space of 15 km2 for isolated systems of a certain
CVE. For example, in Gulf St. Vincent the area of all mangrove patches was
included to calculate the total area of mangrove patches for our study loca-
tions in the Gulf. The same calculation was used for seagrass beds.
To calculate the POM relative contribution to the surface area and stand-
ing biomass of each ecosystem, the percent of surface (% area) was calculated
by dividing the surface area of the ecosystem by the area of the connected
catchment terrestrial system plus all the vegetated intertidal systems
(mangrove patch, seagrass and macroalgal beds and salt marshes). Catch-
ment areas were obtained from nature web maps of rivers or creeks within
watersheds that lead into the respective study area (http://spatialwebapps.
environment.sa.gov.au/naturemaps/?locale=en-us&viewer=naturemaps).
At the west side of Gulf St. Vincent, there is no watershed that could result in
a significant input from terrestrial plants to the Gulf St. Vincent (Fig. 1). Per-
cent of biomass (% biomass) was calculated dividing the biomass of a partic-
ular CVE by the total estimated biomass of all intertidal systems. Biomass of
terrestrial plants of catchment area was not included, as data were not avail-
able. The area- and biomass-based contribution was not calculated for
D. Saavedra-Hortua et al. Science of the Total Environment 896 (2023) 164829

A Connected coastal Isolated mangrove Isolated seagrass

vegetated ecosystems patch bed

Total area (km2) Total area (km2) Total area (km2)

1439.9 2.71 38.1

Isolated Isolated Isolated Isolated

B Connected CVE
mangrove patch seagrass bed C Connected CVE
mangrove patch seagrass bed
3.6
110
AGB (Mg C ha-1)

AGB (Mg N ha-1)

55 1.8
§ §
§ §
* * * *
0 0
Isolated Isolated Isolated Isolated
D Connected CVE
mangrove patch seagrass bed E Connected CVE
mangrove patch seagrass bed
0.30
C stock AGB (Tg C)

N stock AGB (Tg N)

0.01

0.15 0.005

0 0
Mangrove patches Seagrass beds Salt marshes Macroalgae beds

Fig. 2. Surface area and carbon and nitrogen stocks (of standing biomass) for the different coastal vegetated ecosystems (mangrove patches, seagrass beds, macroalgal beds
and salt marshes; see figure legend for details) in the seascape of connected mangrove forest and seagrass beds (Gulf St. Vincent), isolated mangrove patch (Rushy Island), and
isolated seagrass bed (Victor harbor). The pie charts show the percentage occupied by each CVE and the total area occupied by all CVE in km2 (panel A). Carbon (C) and
nitrogen (N) content in standing biomass per unit area (Mg C ha-1) (panel B and C) in the different CVE; error bars represent standard error. Error bars show standard
error. Total carbon and nitrogen stocks (content multiplied by surface area) in the different CVE (panel D and E). AGB = above-ground biomass, Mg = Megagrams,
Tg = Teragrams, ha = hectare. *Data taken from Serrano et al. (2019), and references therein in supplementary material (Clarke and Jacoby, 1994; Kelleway et al.,
2016; Santini et al., 2019, cited in Serrano et al., 2019). §Data taken from Copertino et al. (2005) and Hill et al. (2015). Values are shown in supplementary Table 2.

oceanic sources, because no representative surface area or biomass for the included to calculate the total area of each ecosystem for our study loca-
‘ocean’ could be calculated. tions in the Gulf. In Victor Harbor and Rushy Island the limits of the CVE
maps were set up based on coastal bathymetric lines that showed potential
%Contribution exchange of material with connected CVE. For example in Victor Harbor
Area Relative Contribution ¼ (1)
%Area limits were set in the lowest bathymetric line that show water movement
between seagrass beds and connected macroalgae beds and Rushy Island
%Contribution limits were determined by the lowest bathymetric line that showed where
Biomass Relative Contribution ¼ (2)
%Biomass water movement can occur between the mangrove patch and salt marshes.

In Gulf St. Vincent we assumed that when ebb currents flush the tide out
from all CVE, the POM accumulates in the centre of the Gulf and is mixed.
Afterwards, when the floods inundate the CVE again, the water movement
brings back POM from the centre of the Gulf into the coastal ecosystems.
Therefore, we used the whole area or standing biomass of the CVE inside
the Gulf for a specific CVE. For instance, in Gulf St. Vincent the area of all
mangrove patches, seagrass beds, salt marshes and macroalgal beds were
3. Results
3.1. Carbon and nitrogen content in connected vs isolated mangroves and
seagrass beds
In general, connected seagrass beds had the highest carbon and nitrogen
stocks in their standing biomass compared to other CVE, mainly due to their

6
D. Saavedra-Hortua et al.
Fig. 3. Box-plot of sediment percentages of carbon and nitrogen percentages (panel A and B) and carbon and nitrogen content (panel C and D) shown separately for mangrove
patches (green) and seagrass beds (orange). Carbon and nitrogen content in sediments was taken from the upper 15 cm of the sediments. The thick line inside the box-plot
indicates the median. Error bars show standard error, upper fence of the bars shows maximum values whilst lower fence shows minimum values. Dot points outside boxplots
indicate outliers and different letters indicate significant differences across the ecosystems (Post-hoc Tukey test P < 0.05).
larger area (Fig. 2 Panel A, D, and F). Mangrove patches and seagrass beds
showed variable carbon and nitrogen within standing biomass, depending
on whether the ecosystems were connected or isolated. Connected man-
grove patches had 6–7 times greater carbon and nitrogen content in their
standing biomass per hectare than the isolated mangrove patch (P < 0.05;
Fig. 2 Panel B and C, Supplementary Tables 2 and 3). The opposite was
found for isolated seagrass beds, which had four times higher carbon and
nitrogen in the standing biomass per hectare compared to connected
seagrass beds (Fig. 2 Panel B and C).
Connected mangrove patches had the highest sediment carbon % com-
pared to other ecosystems (P < 0.05, Fig. 3 Panel A and C, Supplementary
Tables 2, 3 and 4). Carbon content was 3 times higher in the sediment of
isolated seagrass compared to the isolated mangrove patch (P < 0.05,
Fig. 3 Panel C). Isolated seagrass beds had significantly higher sediment
carbon content than connected beds (P < 0.05, Fig. 3 and Supplementary
Table 3 and 4).
On average, the nitrogen % and content in the sediment were signifi-
cantly higher (P < 0.05, 3–12 times) in connected mangrove patches com-
pared with seagrass bed sites (both connected and isolated) (Fig. 3 Panel B
and D, Supplementary Tables 3 and 4). Significantly greater nitrogen con-
tent was also found between the isolated mangrove patch and both isolated
and connected seagrass sites (isolated and connected) (P < 0.05, Fig. 3
Panel B and D, Supplementary Tables 3 and 4).
3.2. Contributions of the main primary producers to particulate organic matter:
area and biomass indices
All sediment samples signatures where within the range of the primary
producer used in the MixSIAR model except for samples of suspended par-
ticulate matter in a connected site (Port Wakefield) and sediment traps in
the isolated seagrass site (Stansbury). The δ15N signals of primary pro-
ducers of a connected site (Torrens Island) were higher than those of the
other locations and were, thus, separated from other connected and isolated
seagrasses locations inside Gulf St. Vincent (Supplementary Fig. 1). There-
fore, MixSIAR models were run separately for these locations.
The main primary producers that contributed to suspended particulate
matter and sediment traps in connected CVE were macroalgal thalli,
7
Science of the Total Environment 896 (2023) 164829
mangrove trees and seagrass plants (Fig. 4). In all connected locations,
macroalgal thalli were the greatest contributor (27–50 %), the second or
third biggest contributor were mangrove trees (20–38 %) or seagrass plants
(12–25 %). Oceanic sources were the second biggest contributor only in
Port Clinton (10–22 %) connected location (Fig. 4 and Supplementary
Table 5).
In connected locations, even though salt marshes, macroalgal beds and
mangrove trees had smaller total surface areas and total above ground bio-
mass compared with seagrass beds, these three CVE contributed relatively
highly to suspended particulate matter and trap samples when taken into
account the ecosystems area and standing biomass (Fig. 5). Across the
three connected locations, seagrass beds contributed between 9 and 25 %
to the POM, which was lower compared with macroalgal beds (27–46 %)
and mangrove patches (5–38 %) (Fig. 4 and supplementary Table 4).
When taking into account surface area of the ecosystem, mangrove patches
had area relative indices between 0.1 and 0.7 and macroalgal thalli from
0.3 to 0.6, which were larger compared with that of seagrass beds
(0.01–0.02) (Fig. 5 and Supplementary Table 6). With respect to the stand-
ing biomass, salt marshes presented biomass relative indices between 0.05
and 0.01, that were higher than seagrass beds indices (0.001–0.004), show-
ing that in terms of standing biomass salt marshes contributed to the POM
relatively highly compared with seagrass beds (Fig. 5 and Supplementary
Table 6).
In isolated seagrass beds, seagrass plants, macroalgal thalli and oceanic
sources contributed the most across the isolated locations. In the isolated
seagrass beds of Stansbury the greatest contributors were seagrass plants
(47–85 %), followed by macroalgal thalli (11–28 %) (Fig. 4). These two
sources combined accounted for >70 % of suspended particulate matter
and 90 % of sediment trap samples (Fig. 4). In the other isolated seagrass
beds (Victor Harbor), seagrass plants were again the main contributors
(37–65 %), but in comparison oceanic sources contributed the second
highest (33–35 %), representing >70 % in both suspended particulate mat-
ter and sediment trap samples (Fig. 4 and Supplementary Table 5). When
taking into account the area of the ecosystem and the ecosystem standing
biomass, Victor Harbor's seagrass plants were the greatest contributors
with area indices ranging between 0.6 and 0.7 and biomass indexes be-
tween 0.4 and 0.5. In Stansbury, macroalgal thalli had the largest
D. Saavedra-Hortua et al. Science of the Total Environment 896 (2023) 164829

Fig. 4. MixSIAR model results for connected mangrove patches and seagrass beds and isolated sites. Blue circles ( ) indicate connected mangrove patches and seagrass beds
(Port Clinton, Port Wakefield and Torrens Island), the yellow circles ( ) indicate seagrass beds with no connected mangrove patches (Victor Harbor and Stansbury), and the
light green circle ( ) indicates a mangrove patch with no connected seagrass bed (Rushy Island). Each small figure represents contributions of the different sources. Sediment
traps are depicted on the left side, suspended particulate matter on the right side. At Torrens Island, the figure corresponds only to suspended particulate matter samples,
because trap samples were lost. Contributions are expressed on the y-axis in proportion to 1, where 1 is 100 % of the contribution. See legend for details of the colours
representing the sources. The x-axis shows the transect points across each seascape, with the higher the numbers being the closer to the terrestrial zone. Areas occupied
by different coastal vegetated ecosystems, catchment area and oceanic are shown in different colours (see legend for details).

contribution, with the area and biomass indices ranging between 0.1 and
0.4 in both suspended particulate matter and sediment traps (Fig. 5).
In the isolated mangrove patches, salt marshes were the largest contributor
(16–47 %), followed by mangrove trees (12–27 %), combined both accounted
for >65 % in suspended particulate matter samples and 30 % in trap samples
(Fig. 4 and Supplementary Table 5). Conversely, when taking into account
mangrove patch size and standing biomass, mangrove trees were the greatest
contributors, accounting for >95 % in both suspended particulate matter and
sediment traps (Fig. 5 and Supplementary Table 6). Area and biomass indices
of mangrove patches were between 100 and 250 times larger compared with
that of salt marshes (Fig. 5 and Supplementary Table 6).
4. Discussion
4.1. Carbon and nitrogen content in connected vs isolated mangroves and
seagrass beds
Connected mangrove patches and seagrass beds had higher carbon
stocks in their standing biomass and sediments compared to the isolated
mangrove patch. However, isolated seagrass beds had more carbon content
in the biomass and sediments than seagrass beds which were connected to
mangrove patches. Connectivity between mangrove patches and seagrass
beds can increase carbon content in mangroves by increasing primary pro-
duction (Huxham et al., 2018) and by accumulating more allochthonous
(i.e., seagrass leaves) particulate matter that is sequestered in the sediments
via root-reducing hydrodynamics that allow for particle settlement (Alongi,
2014; Gillis et al., 2014b). Comparing isolated versus connected seagrass
beds, this study found that isolated beds had higher biomass per hectare.
This suggests more carbon is produced by the plants per unit area and
retained in the sediment due to higher belowground biomass production.
Higher above ground biomass may also enhance particle settlement as re-
ported in other studies (Tanaya et al., 2018). Due to localized characteris-
tics such hydrodynamics, sediment fluxes and deposition there is not
always a consistent pattern of sediment density related with carbon and ni-
trogen sediments stocks in coastal ecosystems, which could vary across the
ecosystems as well as the depth of the sediment (Howard et al., 2014; Dahl
et al., 2016; Reis et al., 2017). Isolated seagrass beds had denser sediments
compared with connected seagrass beds and mangrove patches (both

8
D. Saavedra-Hortua et al. Science of the Total Environment 896 (2023) 164829

Area relative contribution Biomass relative contribution

Port 0.08
1
Clinton
0.5 0.04
(Connected
0 0
location) P5 P4 P3 P2 P1 P5 P4 P3 P2 P1 P5 P4 P3 P2 P1 P5 P4 P3 P2 P1

Port
1.2 0.07
Wakefield
0.6 0.035
(Connected
0 0
location) P4 P3 P1 P5 P4 P3 P2 P1 P4 P3 P1 P5 P4 P3 P2 P1

Torrens 1.2 0.08

Island
0.6 0.04
(Connected
0 0
location) P4 P3 P2 P1 P4 P3 P2 P1

0.6 0.04
Stansbury
0.3 0.02

y
(Isolated seagrass
0 0
P4 P3 P2 P1 P4 P3 P2 P1 P4 P3 P2 P1 P4 P3 P2 P1
bed)

Victor 0.75 0.28

Harbor
0.325 0.14
(Isolated seagrass
0 0
bed) P2 P1 P4 P3 P2 P1 P2 P1 P4 P3 P2 P1

Rushy 270 0.12

Island (Isolated
135 0.06
mangrove patch)
0 0
P3 P2 P1 P4 P3 P2 P1 P3 P2 P1 P4 P3 P2 P1

Fig. 5. Area relative contribution and biomass relative contribution of the main coastal vegetated ecosystems mangrove patches, seagrass beds, macroalgal, salt marshes and
terrestrial ecosystems within the coastal catchment area (see legend for details). Each figure represents contributions of the different sources, sediment traps are depicted on
the left side, suspended particulate matter on the right side. In Torrens Islands, the figure corresponds only to suspended particulate matter samples. Area relative contribution
index on the left side and biomass relative contribution on the right side are expressed on the y-axis, see legend for details of the colours representing the sources. The x-axis
shows the transect points across each seascape, higher numbers being closer to the terrestrial zone of the transect.

connected and isolated) due to these isolated beds had more organic carbon
particles in a smaller volume of sediment compared with other ecosystems.
We conclude that connected mangrove patches have higher carbon, which
could be facilitated by exchange of POM with seagrass beds, whilst for
seagrass beds other variables such as sediment density are more important
to determine carbon content.
Mangrove patches had greater nitrogen content in both standing bio-
mass and sediments compared to other ecosystems whether they were con-
nected or isolated from seagrass beds. The greater plant structure of
mangrove trees compared to salt marsh or seagrass plants resulted in the
former having higher nitrogen content in the standing biomass compared
with the other ecosystems per unit area. Higher standing biomass per hect-
are in isolated beds meant that these beds also had higher nitrogen content
compared with connected beds.
Different to our results of higher nitrogen (%) in the mangrove forests
sediments, in the Red Sea no significant difference in nitrogen (%) in the
sediments between mangrove patches and seagrass beds in a connected lo-
cation was found (Garcias-Bonet et al., 2019). This was due to localized
high carbon/nitrogen ratios in mangrove leaves and sediments, compared
with low carbon/nitrogen ratios in seagrass meadows. Mangrove patches
were the most important CVE in terms of nitrogen content since they can
store the highest amount of nitrogen per surface area compared with
other CVEs, mainly due to trees having higher standing biomass per hectare
compared with other primary producers across all CVE.
Connectivity affects ecosystems services of carbon sequestration, and
nitrogen retention differently in mangrove patches and seagrass beds, con-
nectivity with other ecosystems and localized internal ecosystem processes
could be more significant drivers, especially for seagrass beds. Physiologi-
cal process within each ecosystem such as higher net primary production
of mangrove trees could be linked with higher carbon and nitrogen content
per unit area within the ecosystem as well as in connected ecosystems
(Walton et al., 2014; Huxham et al., 2018). However, localized studies com-
paring primary production and export of POM of different sizes across dif-
ferent coastal ecosystems are necessary to elucidate the role of plant
physiological processes for carbon and nitrogen fluxes. Mangrove patches
connected with seagrass and macroalgae beds are highly effective for two
ecosystems services, carbon sequestration and nitrogen retention simulta-
neously. In seagrass beds localized characteristics such as sediment density

9
D. Saavedra-Hortua et al.
and larger biomass per area were more critical for carbon sequestration and
nitrogen retention rather than connectivity with other ecosystems. In man-
grove patches and seagrass beds, plant biomass production per unit area to-
gether with the exchange of particulate organic matter across CVE within
the seascape are important drivers of ecosystem services, such as carbon se-
questration and nitrogen retention.
4.2. Contribution of coastal vegetated ecosystems to particulate organic matter
Isotopic signatures of all samples were within similar the range of pri-
mary producer δ13C and δ15N reported in the area δ15N values of 3.5 ‰
to 7 ‰ and δ13C values of −25.5 ‰ to −7.8 ‰ (Davenport and Bax,
2002) (Supplementary Fig. 1). Some ecosystems (i.e., macroalgae and
seagrass beds) showed high contributions of POM in terms of percentage,
however, when considering their biomass or areal distribution, their rela-
tive contribution was reduced. For example, seagrass beds appeared to be
the largest donor of particulate organic matter in terms of percentage, how-
ever since they had the highest cover area and biomass across all CVEs, we
found that they did not contribute significantly to the POM compared with
other ecosystems when taken into biomass and area. Seagrass beds in
Thailand have been reported to contribute significantly to POM even
though they have a small area (Gillis et al., 2014b), as well as contributing
nitrogen to mangrove patches (Walton et al., 2014). However, in the pres-
ent study their relative contribution to POM was lower than that of
macroalgal beds, salt marshes and connected mangrove patches. They
should be also considered as donors to isolated mangrove patches since
Posidonia sp. leaves were observed in mangrove patches that did not have
connected seagrass beds, such as Port Wakefield.
Previous studies using isotopes chemical tracers have also shown that
contributions of macroalgal beds to POM in mangrove patches were higher
than those of seagrass plants (Saavedra-Hortua et al., 2019). Macroalgal
beds were generally one of highest contributors to carbon and nitrogen in
POM across all CVE. Macroalgal beds contributed to POM across all locations,
regardless of the distance between the sampling locations to the macroalgal
beds (i.e., short distance in the isolated seagrass bed in Victor harbor or
long distance in connected systems in Gulf St. Vincent. This is due to their
characteristics such as high productivity and the ability of algae to drift and
therefore expand the export of their materials (Ortega et al., 2019). In man-
grove patches and seagrass beds in the Gulf St. Vincent, we observed
Hormosira banksii thalli that were not attached to the sediment. Macroalgal
beds were located mainly in the south of the Gulf, indicating that thalli poten-
tially drift up to 100 km. Differences across the main algal taxonomic groups
Chlorophyta, Phaeophyceae and Rhodophyta should be considered in terms
of drifting and therefore exportation of material (Trevathan-Tackett et al.,
2015; Ortega et al., 2019). Recent findings showed that due to high produc-
tivity and abundant standing biomass, macroalgal beds are an important
source of carbon, partly through exporting considerable amounts of carbon
to the seascape (Duarte et al., 2013b; Hill et al., 2015). Our results suggest
that in addition to carbon exportation, macroalgae beds could be important
contributors of nitrogen to other CVE.
Even though mangrove patches and salt marshes had the smallest sur-
face area they still contributed more POM compared to other ecosystems
such as seagrass beds with larger areas, corroborating the outwelling hy-
pothesis (Odum, 1968; Lee, 1995). This indicates that other ecosystems
are dependent on mangrove patches and salt marshes as a carbon and nitro-
gen source, regardless of their occupied area. We showed in our analysis
that area, standing biomass and particulate attributes i.e., macroalgal
drifting ability, are important parameters when determining the particulate
organic matter donation from individual ecosystems. Our findings showed
that conservation of all CVE is important for maintaining ecosystems
services of carbon sequestration and nitrogen retention, due to all CVE
being important contributors of carbon and nitrogen regardless of the
standing biomass and the area occupied. Nevertheless, additional studies
that analyse and determine the origin of the longer-term carbon and nitro-
gen stocks in the sediments will be critical to determine the role of each
CVE across different seascapes.
10
Science of the Total Environment 896 (2023) 164829
Even though catchment areas of this study area have been reported to
be very arid and carrying low terrigenous input to St. Vincent gulf
(Richardson et al., 2005), this study did not include biomass of terrestrial
plants in the catchment area. Therefore, limiting our conclusions on the
role of terrestrial plants and related ecosystems as donors of POM to coastal
ecosystems.
5. Conclusion
Ecosystem connectivity may drive some ecosystem services of global
relevance such as carbon sequestration or nitrogen retention differently in
mangrove patches and seagrass beds. For example, mangrove patches that
were connected to seagrass beds presented higher carbon per unit area in
the sediments compared with isolated patches, although nitrogen retention
does not seem to be affected by connectivity between mangrove patches
and seagrass beds. Ecosystem and physiological processes directly or
indirectly related to organic matter turnover may be supported by exchange
of organic matter between connected ecosystems.
Our results showed that all CVE contribute to POM-exchange across lo-
cations, irrespective of their area and standing biomass. Hence, the degra-
dation of one of these CVE could have severe consequences for connected
ecosystems at the seascape level, for example the lack of allochthonous
POM that would be potentially retained in the sediments. Subsequent, to
keep carbon mitigation and nitrogen retention at coastal landscape the en-
vironmental management plans should keep in healthy state ecosystems
that storages high amounts of carbon and nitrogen (i.e., seagrass beds), as
well as, connected ecosystems that contribute with allochthonous POM
(i.e., mangrove forests and macroalgae beds) even if they have smaller
area. Considering all the CVE as a continuing system with interchanges
across the seascape, rather than each CVE as a separate system will improve
knowledge of carbon and nitrogen stocks at the seascape scale.
Data availability
Data will be made available on request.
Declaration of competing interest
The authors declare the following financial interests/personal relation-
ships which may be considered as potential competing interests: Daniel
Saavedra Hortua, Lucy Gwen Gillis reports financial support was provided
by German Research Foundation.
Acknowledgements
Sample collection was done under the research permit A26622-1 given
by Department of Environment, Water and Natural Resources (DEWN) of
South Australia. We also thank DEWN for providing the geographical infor-
mation for the cartographic analysis. This study was funded by the DFG
(Deutsche Forschungsgemeinschaft), project Number (GI 121011- 1).
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.scitotenv.2023.164829.
References
Adame, M.F., Lovelock, C.E., 2011. Carbon and nutrient exchange of mangrove forests with
the coastal ocean. Hydrobiologia 663, 23–50. https://doi.org/10.1007/s10750-010-
0554-7.
Alongi, D.M., 2014. Carbon cycling and storage in mangrove forests. Annu. Rev. Mar. Sci. 6,
195–219. https://doi.org/10.1146/annurev-marine-010213-135020.
Bates, D., Mächler, M., Bolker, B.M., Waker, S.C., 2015. Fitting linear mixed-effects models
using lme4 Douglas. J. Stat. Softw. 67. https://doi.org/10.18637/jss.v067.i01.
Bulmer, R.H., Schwendenmann, L., Lundquist, C.J., 2016a. Allometric models for estimating
aboveground biomass, carbon and nitrogen stocks in temperate Avicennia marina forests.
Wetlands 1–8. https://doi.org/10.1007/s13157-016-0793-0.
D. Saavedra-Hortua et al.
Bulmer, R.H., Schwendenmann, L., Lundquist, C.J., 2016b. Carbon and nitrogen stocks and
below-ground allometry in temperate mangroves. Front. Mar. Sci. https://doi.org/10.
3389/fmars.2016.00150.
Chen, G., Azkab, M.H., Chmura, G.L., Chen, S., Sastrosuwondo, P., Ma, Z., Dharmawan,
I.W.E., Yin, X., Chen, B., 2017. Mangroves as a major source of soil carbon storage in ad-
jacent seagrass meadows. Sci. Rep. 1–10. https://doi.org/10.1038/srep42406.
Clarke, P.J., Jacoby, C.A., 1994. Biomass and above-ground productivity of salt-marsh plants
in south-eastern Australia. Mar. Freshw. Res. 45 (8), 1521–1528.
Connolly, R.M., Hindell, J.S., Gorman, D., 2005. Seagrass and epiphytic algae support nutri-
tion of a fisheries species. Mar. Eco. Prog. Ser. 286, 69–79. https://doi.org/10.3354/
meps286069.
Connolly, R.M., Gorman, D., Hindell, J.S., Kildea, T.N., Schlacher, T.A., 2013. High congru-
ence of isotope sewage signals in multiple marine taxa. Mar Poll Bull 71, 152–158.
https://doi.org/10.1016/j.marpolbul.2013.03.021.
Copertino, M., Connell, S.D., Cheshire, A., 2005. The prevalence and production of turf-
forming algae on a temperate subtidal coast. Phycologia 44, 241–248. https://doi.org/
10.2216/0031-8884(2005)44[241:TPAPOT]2.0.CO;2.
Dahl, M., Deyanova, D., Gütschow, S., Asplund, M.E., Lyimo, L.D., Karamfilov, V., Santos, R.,
Björk, M., Gullström, M., 2016. Sediment characteristics as an important factor for reveal-
ing carbon storage in <i>Zostera marina</i> meadows: a comparison of four European
areas. Biogeosci. Discuss., 1–30. http://www.biogeosciences-discuss.net/bg-2016-137/.
Davenport, S.R., Bax, N.J., 2002. A trophic study of a marine ecosystem off southeastern
Australia using stable isotopes of carbon and nitrogen. Can. J. Fish Aquat. Sci. 59,
514–530. http://hdl.handle.net/102.100.100/199110?index=1.
de los Santos, C.B., Ege, L.G., Martins, M., Santos, R., Masqué, P., Peralta, G., Brun, F.,
Jiménez-Ramos, R., 2022. Sedimentary organic carbon and nitrogen sequestration across
a vertical gradient on a temperate wetland seascape including salt marshes, seagrass
meadows and rhizophytic macroalgae beds. Ecosystems. https://doi.org/10.1007/
s10021-022-00801-5.
de Vries, F.T., Bardgett, R.D., 2016. Plant community controls on short-term ecosystem
nitrogen retention. New Phytol. 210, 861–874. https://doi.org/10.1111/nph.
13832.
Department for Environment and Water of South Australia (DEW), ). http://spatialwebapps.
environment.sa.gov.au/naturemaps/ Visited on 15–19 Dec. 2019).
Duarte, C.M., Middelburg, J.J., Caraco, N., 2005. Major role of marine vegetation on the oce-
anic carbon cycle. Biogeosciences 2, 1–8. https://doi.org/10.5194/bg-2-1-2005.
Duarte, C.M., Kennedy, H., Marbà, N., Hendriks, I., 2013a. Assessing the capacity of seagrass
meadows for carbon burial: current limitations and future strategies. Ocean Coast.
Manag. 83, 32–38. https://doi.org/10.1016/j.ocecoaman.2011.09.001.
Duarte, C.M., Losada, I.J., Hendriks, I.E., Mazarrasa, I., Marba, N., 2013b. The role of coastal
plant communities for climate change mitigation and adaptation. Nat. Clim. Chan. 3,
961–968. https://doi.org/10.1038/nclimate1970.
Edyvane, K.S., 1996. Marine Biogeography and Conservation Values of the Gulf St. Vincent,
Fleurieu Peninsula Region, South Australia. Technical Report for the Coast Protection
Board. South Australian Research and Development Institute (Aquatic Sciences), West
Beach, Adelaide.
Edyvane, K.S., 1999. Coastal and marine wetlands in Gulf St. Vincent, South Australia: under-
standing their loss and degradation. Wetl. Ecol. Manag. 7, 83–104.
Garcias-Bonet, N., Delgado-Huertas, A., Carrillo-de-Albornoz, P., Anton, A., Almahasheer, H.,
Marbà, N., Hendriks, I.E., Krause-Jensen, D., Duarte, C.M., 2019. Carbon and nitrogen
concentrations, stocks, and isotopic compositions in Red Sea seagrass and mangrove sed-
iments. Front. Mar. Sci. 6, 1–12. https://doi.org/10.3389/fmars.2019.00267.
Gillis, L.G., Bouma, T.J., Jones, C.G., Van Katwijk, M.M., Nagelkerken, I., Jeuken, C.J.L.,
Herman, P.M.J., Ziegler, A.D., 2014a. Potential for landscape-scale positive interactions
among tropical marine ecosystems. Mar. Ecol. Prog. Ser. 503, 289–303. https://doi.
org/10.3354/meps10716.
Gillis, L.G., Ziegler, A.D., Van Oevelen, D., Cathalot, C., Herman, P.M.J., Wolters, J.W.,
Bouma, T.J., 2014b. Tiny is mighty: seagrass beds have a large role in the export of or-
ganic material in the tropical coastal zone. PLoS One 9. https://doi.org/10.1371/jour-
nal.pone.0111847.
Güllstrom, M., Lyimo, L.D., Dahl, M., Eggertsen, M., Anderberg, E., Rasmusson, L.M., Mtwana,
L., Linderholm, H.W., Knudby, A., Bjo, M., 2017. Blue carbon storage in tropical seagrass
meadows relates to carbonate stock dynamics, plant – sediment processes, and landscape
context: insights from the Western Indian ocean. 551–66. https://doi.org/10.1007/
s10021-017-0170-8.
Hill, R., Bellgrove, A., Macreadie, P.I., Petrou, K., Beardall, J., Steven, A., Ralph, P.J., 2015.
Can macroalgae contribute to blue carbon? An Australian perspective. 2, 1689–1706.
https://doi.org/10.1002/lno.10128.
Howard, J., Hoyt, S., Isensee, K., Pidgeon, E., Telszewski, M., 2014. Coastal Blue Carbon.
Available in https://www.cifor.org/publications/pdf_files/Books/BMurdiyarso1401.pdf.
Huxham, M., Whitlock, D., Githaiga, M., Dencer-Brown, A., 2018. Carbon in the coastal sea-
scape: how interactions between mangrove forests, seagrass meadows and tidal marshes
influence carbon storage. Curr. For. Rep. 4, 101–110. http://link.springer.com/10.1007/
s40725-018-0077-4.
Hyndes, G.A., Nagelkerken, I., Mcleod, R.J., Connolly, R.M., Lavery, P.S., Vanderklift, M.A.,
2014. Mechanisms and ecological role of carbon transfer within coastal seascapes. Biol.
Rev. 89, 232–254. https://doi.org/10.1111/brv.12055.
Jennerjahn, T.C., 2020. Relevance and magnitude of ‘blue carbon’ storage in mangrove sedi-
ments: carbon accumulation rates vs. stocks, sources vs. sinks. Estuar. Coast. Shelf Sci.
247, 107027. https://doi.org/10.1016/j.ecss.2020.107027.
Kelleway, J.J., Saintilan, N., Macreadie, P.I., Skilbeck, C.G., Zawadzki, A., 2016. Seventy years
of continuous encroachment substantially increases ‘blue carbon’ capacity as mangroves
replace intertidal salt marshes. Glob. Chang. Biol. 22 (3), 1097–1109.
Kennedy, H., Beggins, J., Duarte, C.M., Fourqurean, J.W., Holmer, M., Marbá, N., Middelburg,
J.J., 2010. Seagrass sediments as a global carbon sink: isotopic constraints. Glob.
Biogeochem. Cycles 24, 1–8. https://doi.org/10.1029/2010GB003848.
11
Science of the Total Environment 896 (2023) 164829
Kindeberg, T., Ørberg, S.B., Röhr, M.E., Holmer, M., Krause-Jensen, D., 2018. Sediment stocks
of carbon, nitrogen, and phosphorus in Danish eelgrass meadows. Front. Mar. Sci. 5,
1–14. https://doi.org/10.3389/fmars.2018.00474.
Krause-Jensen, D., Duarte, C.M., 2016. Substantial role of macroalgae in marine carbon se-
questration. Nat. Geosci. 9, 737–742. https://doi.org/10.1038/ngeo2790.
Krause-jensen, D., Lavery, P., Serrano, O., Masque, P., Duarte, C.M., 2018. Sequestration of
macroalgal carbon: the elephant in the Blue Carbon room. Biol. Lett. 14, 20180236.
https://doi.org/10.1098/rsbl.2018.0236.
Lee, S.Y., 1995. Mangrove outwelling: a review. Hydrobiologia 295, 203–212. https://doi.
org/10.1007/BF00029127.
McGlathery, K., Sundbäck, K., Anderson, I., 2007. Eutrophication in shallow coastal bays and
lagoons: the role of plants in the coastal filter. Mar. Ecol. Prog. Ser. 348, 1–18. https://
www.int-res.com/abstracts/meps/v348/p1-18/.
McLeod, E., Chmura, G.L., Bouillon, S., Salm, R., Björk, M., Duarte, C.M., Lovelock, C.E.,
Schlesinger, W.H., Silliman, B.R., 2011. A blueprint for blue carbon: toward an improved
understanding of the role of vegetated coastal habitats in sequestering CO2. Front. Ecol.
Environ. 9, 552–560. https://doi.org/10.1890/110004.
Mitsch, W.J., Zhang, L., Waletzo, E., Bernal, B., 2014. Validation of the ecosystem services of
created wetlands: two decades of plant succession, nutrient retention, and carbon seques-
tration in experimental riverine marshes. Ecol. Eng. 72 (11), 24. https://doi.org/10.
1016/j.ecoleng.2014.09.108.
Nagelkerken, I., 2009. Ecological Connectivity Among Tropical Coastal Ecosystems. https://
doi.org/10.1007/978-90-481-2406-0_1.
Odum, E.P., 1968. Tidal marshes as outwelling/pulsing systems. In: Weinstein, M.P., Kreeger,
D.A. (Eds.), Concepts and Controversies in Tidal Marsh Ecology. Springer Netherlands,
Dordrecht, pp. 3–7. https://doi.org/10.1007/0-306-47534-0_1.
Ortega, A., Geraldi, N.R., Alam, I., Kamau, A.A., Acinas, S.G., Logares, R., Gasol, J.M.,
Massana, R., Krause-Jensen, D., Duarte, C.M., 2019. Important contribution of
macroalgae to oceanic carbon sequestration. Nat. Geosci. 12, 748–754. https://doi.org/
10.1038/s41561-019-0421-8.
Pei, L., Ye, S., Yuan, H., Pei, S., Xie, S., Wang, J., 2020. Glomalin-related soil protein distribu-
tions in the wetlands of the Liaohe Delta, Northeast China: implications for carbon se-
questration and mineral weathering of coastal wetlands. Limnol. Oceanogr. 65,
979–991. https://doi.org/10.1002/lno.11364.
Phang, V.X.H., Chou, L.M., Friess, D.A., 2015. Ecosystem carbon stocks across a tropical inter-
tidal habitat mosaic of mangrove forest, seagrass meadow, mudflat and sandbar. Earth
Surf. Process. Landforms 40, 1387–1400. https://doi.org/10.1002/esp.3745.
Phillips, J.A., 2001. Marine macroalgal biodiversity hotspots: why is there high species rich-
ness and endemism in southern Australian marine benthic flora? Biodivers. Conserv.
10, 1555–1577. https://doi.org/10.1023/A:1011813627613.
R Core Team, 2013. R: A language and environment for statistical computing. R Foundation
for Statistical Computing, Vienna, Austria. http://www.R-project.org/.
Reis, C.R.G., Nardoto, G.B., Oliveira, R.S., 2017. Global overview on nitrogen dynamics in
mangroves and consequences of increasing nitrogen availability for these systems. Plant
Soil 410, 1–19. https://doi.org/10.1007/s11104-016-3123-7.
Richardson, L., Mathews, E., Heap, A., 2005. Geomorphology and sedimentology of the South
Western planning area of Australia: review and synthesis of relevant literature in support
of regional marine planning. Geosci. Aust. Record 2005/17. 124pp. ISBN: 1 920871 56 X
GeoCat No. 63721. Page range 74-85.
Rogers, K., Macreadie, P.I., Kelleway, J.J., Saintilan, N., 2019. Blue carbon in coastal land-
scapes: a spatial framework for assessment of stocks and additionality. Sustain. Sci. 14,
453–467. https://doi.org/10.1007/s11625-018-0575-0.
Saavedra-Hortua, D.A., Zimmer, M., Gillis, L.G., 2019. Sources of particulate organic matter
across mangrove forests and adjacent ecosystems in different geomorphic settings. Wet-
lands 40, 1047–1059 (2020). https://doi.org/10.1007/s13157-019-01261-9 (2020).
Saderne, V., Geraldi, N.R., Macreadie, P.I., Maher, D.T., Middelburg, J.J., Serrano, O.,
Almahasheer, H., Arias-Ortiz, A., Cusack, M., Eyre, B.D., Fourqurean, J.W., Kennedy,
H., Krause-Jensen, D., Kuwae, T., Lavery, P.S., Lovelock, C.E., Marba, N., Masqué, P.,
Mateo, M.A., Mazarrasa, I., McGlathery, K.J., Oreska, M.P.J., Sanders, C.J., Santos, I.R.,
Smoak, J.M., Tanaya, T., Watanabe, K., Duarte, C.M., 2019. Role of carbonate burial in
Blue Carbon budgets. Nat Commun 10, 1106. https://doi.org/10.1038/s41467-019-
08842-6.
Saderne, V., Cusack, M., Serrano, O., Almahasheer, H., Krishnakumar, P.K., Rabaoui, L.,
Qurban, M.A., Duarte, C.M., 2020. Role of vegetated coastal ecosystems as nitrogen
and phosphorous filters and sinks in the coasts of Saudi Arabia role of vegetated coastal
ecosystems as nitrogen and phosphorous fi lters and sinks in the coasts of Saudi Arabia.
Environ. Res. Lett. 15, 34058. https://doi.org/10.1088/1748-9326/ab76da.
Santini, N.S., Lovelock, C.E., Hua, Q., Zawadzki, A., Mazumder, D., Mercer, T.R., ... Thomas,
T., 2019. Natural and regenerated saltmarshes exhibit similar soil and belowground or-
ganic carbon stocks, root production and soil respiration. Ecosystems 1–20.
Santos, R., Duque-Núñez, N., de los Santos, C.B., Martins, M., Carrasco, A.R., Veiga-Pires, C.,
2019. Superficial sedimentary stocks and sources of carbon and nitrogen in coastal vege-
tated assemblages along a flow gradient. Sci. Rep. 9, 610. https://doi.org/10.1038/
s41598-018-37031-6.
Saunders, D.L., Kalff, J., 2001. Nitrogen retention in wetlands, lakes and rivers. Hydrobiologia
443 (1–3), 205–212. https://doi.org/10.1023/A:1017506914063.
Serrano, O., Ruhon, R., Lavery, P.S., Kendrick, G.A., Hickey, S., Masqué, P., Arias-Ortiz, A.,
Steven, A., Duarte, C.M., 2016. Impact of mooring activities on carbon stocks in seagrass
meadows. Sci. Rep. 6, 23193. http://www.nature.com/articles/srep23193.
Serrano, O., Lovelock, C.E., Atwood T, B., Macreadie, P.I., Canto, R., Phinn, S., Arias-Ortiz, A.,
Bai, L., Baldock, J., Bedulli, C., Carnell, P., Connolly, R.M., Donaldson, P., Esteban, A.,
Ewers Lewis, C.J., Eyre, B.D., Hayes, M.A., Horwitz, P., Hutley, L.B., CRJ, Kavazos,
Kelleway, J.J., Kendrick, G.A., Kilminster, K., Lafratta, A., Lee, S., Lavery, P.S., Maher,
D.T., Marbà, N., Masque, P., Mateo, M.A., Mount, R., Ralph, P.J., Roelfsema, C.,
Rozaimi, M., Ruhon, R., Salinas, C., Samper-Villarreal, J., Sanderman, J.J., Sanders, C.,
Santos, I., Sharples, C., ADL, Steven, Cannard, T., Trevathan-Tackett, S.M., Duarte,
D. Saavedra-Hortua et al.
C.M., 2019. Australian vegetated coastal ecosystems as global hotspots for climate change
mitigation. Nat Commun 10, 4313. https://doi.org/10.1038/s41467-019-12176-8.
Stock, B.C., Jackson, A.L., Ward, E.J., Parnell, A.C., Phillips, D.L., Semmens, B.X., 2018. Ana-
lyzing mixing systems using a new generation of Bayesian tracer mixing models. PeerJ 6,
e5096. https://doi.org/10.7717/peerj.5096.
Tanaya, T., Watanabe, K., Yamamoto, S., Hongo, C., Kayanne, H., Kuwae, T., 2018. Contribu-
tions of the direct supply of belowground seagrass detritus and trapping of suspended or-
ganic matter to the sedimentary organic carbon stock in seagrass meadows.
Biogeosciences 15, 4033–4045. https://www.biogeosciences.net/15/4033/2018/.
Trevathan-Tackett, S.M., Kelleway, J., Macreadie, P.I., Beardall, J., Ralph, P., Bellgrove, A.,
2015. Comparison of marine macrophytes for their contributions to blue carbon seques-
tration. Ecology 96, 3043–3057. https://esajournals.onlinelibrary.wiley.com/doi/abs/
10.1890/15-0149.1.
Valiela, I., Cole, M.L., 2002. Comparative evidence that salt marshes and mangroves may pro-
tect seagrass meadows from land-derived nitrogen loads. Ecosystems 5, 92–102. https://
doi.org/10.1007/s10021-001-0058-4.
Walton, M.E.M., Al-Maslamani, I., Skov, M.W., Al-Shaikh, I., Al-Ansari, I.S., Kennedy, H.A., Le
Vay, L., 2014. Outwelling from arid mangrove systems is sustained by inwelling of
12
Science of the Total Environment 896 (2023) 164829
seagrass productivity. Mar. Ecol. Prog. Ser. 507, 125–137. https://doi.org/10.3354/
meps1082.
Willoughby, N., Thompson, D., Royal, M., Miles, M., 2018. South Australian land cover layers:
An Introduction and Summary Statistics. South Australian Department of Environment
and Water, Adelaide, South Australia, p. 115.
Woodroffe, C.D., Rogers, K., McKee, K.L., Lovelock, C.E., Mendelssohn, I., a., Saintilan N.,
2016. Mangrove sedimentation and response to relative sea-level rise. Annu. Rev. Mar.
Sci. 8 annurev-marine-122414-034025. http://www.annualreviews.org/doi/10.1146/
annurev-marine-122414-034025 annurev-marine-122414-034025.
Xu, H., Qu, Q., Li, P., Guo, Z., Wulan, E., Xue, S., 2019. Stocks and stoichiometry of soil or-
ganic carbon, total nitrogen, and total phosphorus after vegetation restoration in the
Loess Hilly Region, China. Forests 10 (1), 27. https://doi.org/10.3390/f10010027.
Zuur, A.F., Ieno, E.N., Walker, N.J., Saveliev, A.A., Smith, G.M., 2009. In: Zuur, A.F., Ieno,
E.N., Walker, N., Saveliev, A.A., Smith, G.M. (Eds.), Things are not always linear; additive
modelling BT - mixed effects models and extensions in ecology with R. Springer
New York, NY, pp. 35–69. https://doi.org/10.1007/978-0-387-87458-6_3.