Trophic connectivity between the terrestrial and marine ecosystems of Malpelo Island, Colombia, evaluated through stable isotope analysis

The trophic dynamics of islands with low terrestrial primary productivity often depend on marine allochthonous inputs from nearby donor habitats. For instance, on Malpelo Island, Colombia (4° 00′ 05.63″ N; 81° 36′ 36.41″ W), the Nazca booby Sula granti affects the productivity and trophic dynamics of the terrestrial ecosystem by delivering nutrients, primary in the form guano, chick carcasses, and eggs. This study evaluated the trophic connectivity between the terrestrial and marine ecosystems of Malpelo Island, Colombia based on the isotopic (δ13C and δ15N) assessment of 403 samples (107 terrestrial and 296 marine) collected between 2017 and 2021. Isospaces were estimated based on δ13C and δ15N values, contribution of terrestrial sources in consumer diets (mixing model), 15 N enrichment in C3 plants, and interactions among environments (overlap). δ13C and δ15N values showed a larger terrestrial isospace (134.7‰2) than the marine isospace (117.2‰2). The mixing model indicated that detritusTerrestrial (median: 30.2%) contributed more to the food web than C3 plants (0.2%), reflecting high δ13CMarine content. The high isotopic overlap (> 60%) between terrestrial and marine isospaces suggests a significant trophic connection between environments. These results show the role of the marine ecosystem on the terrestrial ecosystem and the importance of S. granti regarding nutrient transfer between environments. The conservation of this seabird is essential to maintain the balance of this insular ecosystem. Using stable isotopes, this study was able to reveal trophic relationships between ecosystems associated with small oceanic islands that host large seabird colonies but have low primary productivity.


Introduction
Organisms can act as biological vectors in the transfer of nutrients among ecosystems (Macavoy et al. 2009;Qin et al. 2014). Thus, allochthonous inputs from different sources (e.g., detritus, food remains) can influence energy, carbon, and nutrient reservoirs (e.g., nitrogen and phosphorus) (Polis and Hurd 1996a;Polis et al. 1997b). In this sense, a key element in the trophic dynamics of islands and coastal areas is a subsidy from a donor habitat through marine allochthonous inputs (Polis and Hurd 1996a;Polis et al. 1997a). Although islands can have low terrestrial primary productivity (Caut et al. 2012), they can support a high abundance and biomass (i.e., secondary production) of consumers subsidized by marine contributions (Sánchez-Piñero and Polis 2000; Moore et al. 2004). These are mainly incorporated from two sources: (1) marine detritus transported across beaches and (2) seabird colonies Hurd 1995, 1996b). This contributes to the combating nutrient limitation (e.g., nitrogen and phosphorus) of primary producers in these low productivity islands (Polis and Hurd 1996a). Marine birds redistribute nutrients among ecosystems (Hentati-Sundberg et al. 2020) and are considered one of the best transport vectors of marine nutrients into terrestrial environments (Otero et al. 2018;Michelutti et al. 2009). Through this process, the nesting colonies of seabirds impact the local dynamics of terrestrial ecosystems (Otero et al. 2018;Hentati-Sundberg et al. 2020) due to the deposition of high quantities of nutrients and organic matter in the form of guano, food leftovers, eggs, feathers, and carcasses (Polis and Hurd 1996a, b). This increases nutrient availability as a support for local biological production (Hentati-Sundberg et al. 2020), which generates high concentrations of parasites around marine bird colonies (e.g., Acari and Diptera) that are a food source for spiders, scorpions, ants, and lizards Hurd 1995, 1996b).
Some isolated systems, such as oceanic islands, can support relatively complex food webs due to the input of nutrients via seabirds (Polis and Hurd 1996a;Polis et al. 1997a;Ellis 2005). This allows a connection between lowproductivity habitats ("receptor habitats") and environments with higher primary productivity ("donor habitats"). These processes drive the trophic and ecological dynamics of connected ecosystems Hurd 1995, 1996a;Polis and Strong 1996;Polis et al. , 1997aAnderson and Polis 1999;Caut et al. 2012).
The terrestrial ecosystem of the Malpelo Flora and Fauna Sanctuary (FFS) is a small insular system (1.2 km 2 ; Graham 1975) with a limited capacity for atmospheric nitrogen fixation. Considering its topography, as well as the presence of the largest nesting colony of the Nazca booby Sula granti (> 80,000 individuals;López-Victoria and Rozo 2007;García 2013), the structure and trophic dynamics of Malpelo Island could be impacted by the input of marine nutrients denoting (i) high connectivity between ecosystems (Wolda 1975;von Prahl 1990;López-Victoria et al. 2009) and (ii) the important role of S. granti in supplying nitrogen from the marine environment (Wolda 1975;López-Victoria et al. 2009). This seabird provides high quantities of nutrients in the form of guano, feathers, eggs, seafood waste, as well as carcasses of chicks, juveniles, and adults (López-Victoria and Werding 2008;López-Victoria et al. 2009. This highlights the potential importance of S. granti in the transport of nutrients from the marine to the terrestrial ecosystem of Malpelo FFS (López-Victoria et al. 2009).
Variation in the resources available in the marine ecosystem around Malpelo FFS (i.e., donor-controlled habitat; Polis et al. 1997a) could modify the community ecology of Malpelo FFS (Wolda 1975), with drastic consequences on species composition and trophic dynamics at landscape scales (Polis and Hurd 1995;Nakano et al. 1999). Changes in the feeding habits of S. granti could result in modifications of the role that this seabird plays in the trophic connectivity between the two ecosystems (Wolda 1975;von Prahl 1990;López-Victoria et al. 2009). Several studies have indicated that of the total energy contributed by S. granti, 99% corresponds to guano, 0.64% to eggs and chicks, and 0.06% to carcasses (López-Victoria et al. 2009). These are important dietary components of the dotted galliwasp Diploglossus millepunctatus, the Malpelo anole Anolis agassizi, the terrestrial crab Johngarthia malpilensis, as well as other invertebrates (López-Victoria 2006;López-Victoria and Werding 2008;López-Victoria et al. 2011).
Trophic studies based on terrestrial macro-species from Malpelo FFS (López-Victoria 2006;López-Victoria and Werding 2008;López-Victoria et al. 2011, including trophic relationships with S. granti, have shown the importance of this avian species in the trophic dynamics (López-Victoria et al. 2009) and stability of the terrestrial ecosystem due to energy inputs from the sea (Wolda 1975;von Parhl 1990;López-Victoria et al. 2009). However, previous trophic studies [direct observations and stomach contents analysis (SCA)] on several terrestrial species of Malpelo FFS should be complemented with other methods to strengthen the hypothesis that the terrestrial ecosystem depends on inputs from the marine ecosystem (Wolda 1975;von Parhl 1990;López-Victoria et al. 2009). Stable isotope analysis (SIA) is a complementary approach that counters some of the study limitations that provide a temporal snapshot of the ingested food (SCA and direct observations). SIA allows the identification of carbon and nitrogen sources of the food assimilated over the short-and long-term (e.g., . The isotopic signal depends on the trophic level and origin of the diet, ingestion rates, accumulation, turnover rates of assimilated tissues, growth, among other factors (Fry and Arnold 1982;Tieszen et al. 1983).
Four specific objectives were addressed in this study to determine the coupling between the terrestrial and marine environments of Malpelo FFS. First, the identification of primary sources that support the terrestrial food web. Second, the identification of the isotopic similarity between the terrestrial and basal marine organisms as an indicator of marine isotopic signals in the terrestrial ecosystem. Third, inference regarding the potential influence of N in guano in the cover of terrestrial plants. Fourth, estimation of the isotopic space (isospace) from the assessment of δ 13 C and δ 15 N values of the biological components of Malpelo FFS to investigate the trophic interaction (connectivity) between both ecosystems (terrestrial and marine).
The general aim was to determinate whether S. granti is the main mediator in the transfer of matter and energy between both ecosystems (terrestrial and marine) and generate new information to clarify some hypotheses. First, the terrestrial food web has a low dependence (relative contribution) on terrestrial C 3 plants due to their low abundance; therefore, terrestrial detritus (sources that consist of decomposing organic matter [DOM] and seabird feces; López-Victoria et al. 2009) should be the basal source that has the greatest percentage of relative contribution to the diet of the different components of the terrestrial food web. Second, the δ 13 C of terrestrial detritus should be similar or vary slightly in relation to the δ 13 C of basal sources (phytoplankton and algae) and marine consumers with a low trophic level (zooplankton, crustaceans, etc.), as well as S. granti's eggs since detritus is mainly composed of guano from S. granti (López-Victoria et al. 2009). Third, that terrestrial C 3 plants should reflect high 15 N-enrichment due to the high concentrations of N in the guano of S. granti (e.g., isotopic enrichment). Finally, if the terrestrial and marine ecosystems of Malpelo Island have a high interaction and trophic dependence, they should display a high level of isotopic overlap as a result of the high connectivity between them.

Study area
Malpelo Island (Fig. 1A) is the summit of a submarine mountain range called the Malpelo Ridge, which extends in a NE-SW direction; it is approximately 241.4 km long and 80.5 km wide (Fig. 1B, red polygon). The island has a maximum height of 300 m above sea level (m.a.s.l) and emerges from approximately 4000 m of depth (Fig. 1C). It is a small oceanic island (1.2 km 2 ; Graham 1975)   Plan 2015). It is influenced by several marine currents due to its isolation and geographical position (Fig. 1D), which generates high productivity in the marine habitat as a result of annual upwelling that supplies nutrients from deep waters. This upwelling supports a diverse and abundant community of consumers whose growth is reliant on this productivity (Rodríguez-Rubio and Stuardo 2002). Thus, the island is an ideal place for the aggregation of species (endemic and migratory; Management Plan 2015). For these reasons, it is part of the Malpelo FFS, which is the largest marine protected area in the Colombian Pacific ( Despite its physical characteristics (volcanic rock), several species of flora and fauna inhabit the island (Management Plan 2015). There are 31 terrestrial photosynthetic organisms represented by a non-vascular plant (moss Octoblepharum albidum), three identified vascular plants (a fern Pityrogramma calomelanos, a C 4 grass Paspalum sp., and a legume) and a currently unidentified vascular plant (i.e., shrubs), 25 lichen species, and a non-lichenized basidiomycete fungus (Leucicoprinus birnbaumii) (von Prahl 1990;González-Román et al. 2014).
The terrestrial fauna of the island includes ~ 40 species of invertebrates (Wolda 1975;Management Plan 2015), such as ants (Odontomachus bauri), beetles (Platynini sp.), and an endemic decapod crustacean (J. malpilensis), three species of endemic reptiles (A. agassizi, D. millepunctatus, and Phyllodactus transversalis), and a high diversity of migratory and resident birds (> 60 species; Management Plan 2015). The largest S. granti nesting colony in the world is found on Malpelo Island, and the species is the most abundant bird in the area (López-Victoria and Rozo 2007; García 2013).

Collection of samples
Samples of 16 terrestrial and 38 marine species/functional groups (Table 1) were collected between 2017 and 2021 in the Malpelo FFS (Fig. 1A). The detritus (n = 5; sources that consist of decomposing organic matter [DOM] and seabird feces; López-Victoria et al. 2009) and guano (n = 1) were collected in test tubes in 2018 and 2021, respectively. All terrestrial samples were collected in October 2018. Samples of mosses (terrestrial C 3 plants) were collected from small cracks with moist soil. Samples of A. agassizi and D. millepunctatus consisted of 1-2 cm of tissue collected from the posterior portion of the tail. The body feathers of S. granti were taken from the root, and whole eggs (yolk and white) were also collected. For invertebrates such as the land crab J. malpilensis, one of the hind limbs was collected, whereas small invertebrates (i.e., millipedes, isopods, spiders, worms, crickets, and ants; Table 1) were collected whole. All terrestrial components were collected in a 100 m radius from the camp of the Armada Nacional de Colombia.
Marine samples were obtained at different depths (between 10 and 30 m) by scuba diving at different sites around Malpelo Island. The muscle tissue of teleost fishes and rays was obtained with a harpoon and/or Hawaiian hook, as well as from fish that had been illegally caught and seized by the authorities. The muscle tissue of the scalloped hammerhead (Sphyrna lewini) and silky shark (Carcharhinus falciformis) was obtained from Estupiñán-Montaño et al. (2017).
Plankton samples were collected around Malpelo Island with a "bongo" type net of 68, 90, and 294 μm mesh size; surface tows were conducted from the M/N Seawolf inflatable boats for 10 min at each sampling site around the island. Samples of the other marine species/groups (e.g., algae, crustaceans, gastropods, and oysters; Table 1) were collected manually.
All samples collected (terrestrial and marine) were placed in pre-labeled zip-lock plastic bags, except for the plankton samples, which were stored in 250 ml plastic bottles. Samples were kept frozen on board the Pacific Diving Company's M/N Seawolf for subsequent transfer to the laboratory. Sampling procedures were endorsed by Parques Nacionales de Colombia, through Memorandum 20177730007973 of 30 May 2017, issued by the Planning and Management Group.

Sample preparation and analysis
Terrestrial and marine samples were washed with distilled water, placed in an oven to dry at 60 °C for 24 h, and subsequently ground to a fine powder with an agate mortar. Sample size from 0.23 to 0.97 mg and between 0.44 and 3.70 mg of fine powder was obtained from terrestrial and marine specimens, respectively, and posteriorly stored in 3.2 × 4-mm tin capsules.
Some samples could contain high lipid contents which are 13 C-depleted relative to proteins (Newsome et al. 2010;Carlisle et al. 2017) and chitin in pure form (i.e., as polymeric acetylated glucosamine), which is generally crosslinked to proteins (Webb et al. 1998) and may affect δ 13 C values; while urea (NH 2 ) 2 CO can bias δ 15 N values as it is 15 N-depleted (Logan et al. 2008;Carlisle et al. 2017). Therefore, specific treatments are necessary for the extraction of lipids, urea , and chitin (DeNiro and Epstein 1978). However, chemical methods of lipid extraction can also introduce errors when urea and lipids are removed ; whereas the extraction of pure chitin can cause deacetylation (Webb et al. 1998) and contamination by additional proteins that For the above reasons, this study applied mathematical algorithms as a treatment for the correction of the δ 13 C values (except for sharks, and seabird eggs and feathers; see below). The C:N ratio of all samples (terrestrial and marine) was estimated and compared with reference values to determine which sample should be corrected. For the soft tissues (i.e., muscle), the reference value of the C:N ratio was ≤ 3.5, which indicates that there is no effect of lipid content, whereas C:N values > 3.5 suggest a high lipid content (Post et al. 2007). δ 13 C values of terrestrial and marine samples (Table 1) with C:N values > 3.5 were mathematically normalized according to Kiljunen et al. (2006): where δ 13 C adjusted is the δ 13 C after normalization and δ 13 C measured is the δ 13 C obtained from the sample without lipid removal. D and I are constants of the algorithm with values of 7.018 and 0.048, respectively; L is the estimated proportional lipid content of a sample and was estimated as L = − 20.54 + (7.24 × C:N), where − 20.54 and 7.24 are constants of the algorithm and C:N is the ratio of carbon and nitrogen of the sample analyzed (Post et al. 2007).
For the case of arthropods (i.e., ants, isopods, millipedes, and crabs; Table 1), the C:N ratio was compared with the theoretical value of the C:N ratio for pure chitin of 6.86 (Schimmelmann and DeNiro 1986), so that the C:N ratios < 7. 0 were accepted as pure chitin (Webb et al. 1998;Pringle and Fox-Dobbs 2008) and were not mathematically normalized (Schimmelmann and DeNiro 1986;Webb et al. 1998;Pringle and Fox-Dobbs 2008). Otherwise, δ 13 C values were normalized according to Kiljunen et al. (2006) (Eq. 1).
(1) 13 C adjusted = 13 C measured + D × I + 3.90 1 + 287∕L , Sula granti feathers were cleaned of surface lipids and contaminants using a 2:1 chloroform:methanol solution, followed by two successive methanol rinses (Jaeger et al. 2009). The δ 13 C values of S. granti eggs were mathematically normalized because lipid extraction can alter δ 15 N by washing out nitrogenous compounds. In this case, the formula proposed by Elliot et al. (2014) was used: where δ 13 C lipid-extracted is the δ 13 C after normalization and δ 13 C non-extracted is the δ 13 C obtained from the sample without lipid removal.
Finally, the extraction of lipids and urea from elasmobranch muscle samples (i.e., sharks and rays; Table 1) was performed following the procedure described by .
Stable isotope analyses (i.e., carbon and nitrogen) of terrestrial and marine specimens were carried out in the Stable Isotope Laboratory of the Instituto Andaluz de Ciencias de la Tierra in Granada (CSIC-UGR), Spain, using an online Carlo Erba NA 1500 NC elemental analyzer coupled online via ConFlo III interface to a Delta Plus XP mass spectrometer (EA-IRMS; ThermoQuest). Stable isotopes were reported as δ values per mil (‰) based on the following equation: where R is the isotopic ratio ( 13 C/ 12 C or 15 N/ 14 N) of the sample or the standard (V-PDB and AIR for carbon and nitrogen, respectively). Commercial CO 2 and N 2 were used as internal standards for isotopic analyses. Internal standards of − 30.6‰ and − 11.7‰ (V-PDB) were used for δ 13 C analysis and internal standards of − 1.0‰ and + 16.0‰ (2) δ 13 C lipid-extracted = 13 C non-extracted + 1.47 − 2.72 × Log 10 (C:N),  Kiljunen et al. (2006) (AIR) were used for δ 15 N. Standards were systematically interspersed among analytical batches; the daily drift of the mass spectrometer was corrected and a precision factor < ± 0.1‰ for δ 13 C and δ 15 N was calculated. Reference gases and in-house standards (with different C:N ratios and isotopic composition) were calibrated against International Reference Materials for carbon (USGS-24 and IAEA-C6) and nitrogen (IAEA-N1, IAEA-N2, and IAEA-N3).

Relative contribution of terrestrial basal sources to the trophic web
The relative contribution of potential terrestrial basal sources to the diet of terrestrial consumer groups was estimated with the package simmr (version 0.3) in R (version 4.1.2; R Core Team 2018). This model uses a Bayesian isotopic framework based on δ 13 C and δ 15 N values to estimate the proportional contribution of potential prey (in this case, basal sources) to a consumer's diet (Parnell et al. 2013), including variability in model inputs such as trophic discrimination factor (TDF) values of consumers.
The following two steps were implemented to run the mixing model. First, the relative contribution of two potential basal sources, i.e., terrestrial C 3 plants and terrestrial detritus to the diet of terrestrial consumers (i.e., mixtures), was estimated from the δ 13 C values (Supplementary material S1).
The second step was implemented in four parts (Supplementary material S2). First, two potential basal sources (C 3 plants and detritus). Second, all terrestrial consumers were considered potential sources (i.e., prey) and potential consumers (i.e., mixing) due to their feeding behaviors and the high degree of omnivory in the terrestrial ecosystem (López-Victoria 2006; López-Victoria and Werding 2008; López-Victoria et al. 2011). Thus, including all organisms as potential prey allowed us to obtain the terrestrial isotopic space of the Malpelo FFS from which the contribution of the basal sources to the diet of each of them would be estimated. The mixing model functions used (simmr) allowed us to select and compare the contribution of two or more sources to the diet of the consumers by selecting them from the isotopic space. In this case, the contribution value of plants and detritus was obtained for each consumer. Third, due to the lack of specific TDFs for each terrestrial organism, the estimated mean TDF was used for terrestrial ecosystems (Δ 13 C = 0.5 ± 0.19‰ SD and Δ 15 N = 2.3 ± 0.24‰ SD; McCutchan et al. 2003) to minimize sources of uncertainty (i.e., environmental and physiological factors, trophic position, metabolic rates, growth rates; Phillips et al. 2014), to which mixing models are highly sensitive (Phillips et al. 2014). Fourth, the mixing model was adjusted to verify that the TDFs, potential prey, and consumers were consistent with the assumptions of the model (Smith et al. 2013).
The mixing model adjustment was run with 10 3 iterations with a 95% probability for the mixing polygon (Smith et al. 2013). The model was considered adequate if isotopic values were within 1% of the mixing model polygons (Reum et al. 2020). Finally, once the model was correctly adjusted, the mixing model was run with the isotopic values of terrestrial Malpelo FFS consumer groups (Table 1). The mixing model was run as 4 Markov Chain Monte Carlo (MCMC) chains with 10 6 iterations, the first 10 4 were discarded (burn-in period), and the remainder thinned by a factor of 100.
We did not directly analyze sea bird guano samples from the Malpelo FFS. However, if we consider that the terrestrial detritus of the Malpelo FFS is mostly composed of seabird guano (López-Victoria et al. 2009), it can be considered a proxy to identify marine inputs (i.e., origin of basal sources). Therefore, to confirm this hypothesis, the δ 13 C detritus of the terrestrial ecosystem was compared to the δ 13 C values of five marine groups, i.e., macroalgae, phytoplankton, zooplankton, crustaceans, and S. granti eggs as indicators of the food consumed by mothers during egg formation (Hobson 2006). For this purpose, the δ 13 C values of the marine groups (except macroalgae and phytoplankton) and of the terrestrial detritus were corrected with the mean TDF for marine environments (Δ 13 C = 0.4 ± 0.17‰ SD; McCutchan et al. 2003) and statistically compared using a non-parametric paired test (Wilcoxon rank sum test) with a significance level of 0.05.

N-enrichment
15 N-enrichment of terrestrial components was estimated three times using the average δ 15 N values of detritus, S. granti's eggs, and S. granti's feathers as a reference component since they provide marine nutrients to the terrestrial ecosystem (García and López-Victoria 2007;López-Victoria et al. 2009). Relative 15 N enrichment was calculated using the algorithm proposed by Estrada et al. (2006): Y and z are the elements and atomic mass of interest ( 15 N), and x are the terrestrial components (i.e., plants, A. agassizi, J. malpilensis, D. millepunctatus, ants, millipedes, and Isopoda; Table 1). The isotopic enrichment will be estimated based on the reference component (i.e., detritus and S. granti's eggs and feathers).

Isotopic niche and isotopic overlap
To quantify the isotopic niche and isotopic overlap between ecosystems (terrestrial [with and without C 3 plants] vs. (4) marine), the Stable Isotope Bayesian Ellipses package (SIBER, Jackson et al. 2011) available in the R software (R Development Core Team 2021) was used. This analysis estimates ellipses that represent the "core isotopic niche" (Standard Ellipse Corrected Area, SEA C ) using a Bayesian approach and calculating covariance matrices that define the shapes and areas of the ellipses (Jackson et al. 2011). The ellipses (40%) were corrected using a posteriori randomly replicated sequences (SEA C ; Jackson et al. 2011), which represent the isotopic niche width of consumers. In addition, this method allows the estimation of the isotopic niche overlap of the consumer (in this study, isospace) based on the overlap between ellipses (Newsome 2007). SIBER results were supported by the nicheROVER package in R (Lysy et al. 2014), which is a Bayesian method that calculates the probability of overlap between niches pairwise using multidimensional niche indicator data. The probabilistic density of niche overlap was calculated by running 10 4 iterations and 95% probability of the data from each isospace, with the advantage that the package provides directional niche overlap estimates (x vs y and y vs x), according to the distributions of a specific species in the multivariate niche space (Lysy et al. 2014).

Results
A total of 403 samples were collected in Malpelo FFS, of which 27% (n = 107) corresponded to the terrestrial ecosystem and 73% (n = 296) to the marine ecosystem (Table 1). δ 13 C values of the terrestrial ecosystem ranged from − 30.3 to − 15.0‰ and δ 15 N ranged from 3.7 to 21.3‰ (Table 1). Terrestrial C 3 plants (moss Octoblepharum albidum) had the lowest average δ 13 C value (− 30.3‰) and the dotted galliwasp Diploglossus millepunctatus had the highest average value (− 15.0‰), with a total range in δ 13 C values of 15.3 ‰ (Fig. 2). The lowest δ 15 N value corresponded to the terrestrial C 3 plants (3.7 ‰), whereas the highest value was obtained for arthropods from the family Araneae (21.3 ‰), with a δ 15 N range of 17.6 ‰ (Fig. 2).

Relative contribution of terrestrial basal sources to the trophic web
The fitted model of the mixing polygons and their subsequent predictive validations suggested that these results explained the uncertainty of the TDFs and isotopic values of the 13 consumer groups (Supplementary material S3A). Therefore, the implementation of the mixing model was adequate to estimate the relative contribution of the different basal sources, confirmed by the Gelman-Rubin (Rhat) convergence diagnostic statistics, which was 1.00 for all parameters and suggested that there was convergence. The estimated relative contribution of the two terrestrial sources in the trophic web indicated that the main δ 13 C supply to the trophic web was provided by detritus (median [95% credibility limits]: 98.8% [96.5-99.7%]) concerning terrestrial C 3 plants (1.2% [0.3-3.5%]) (Fig. 4). These results suggest that the lizard Anolis agassizi, the crab Johngarthia Fig. 3 Marine isospace of the Malpelo Fauna and Flora Sanctuary, Colombia, represented as average values (± SD) of δ 13 C and δ 15 N for 39 different consumer groups (species/families/orders) of the marine trophic web Fig. 4 A Plot of the terrestrial basal sources (mean ± SD) and consumers (orange points) used to run the stable isotope mixing models with one isotope (i.e., δ 13 C). B Estimation of the contribution prob-ability (in %) of the terrestrial basal sources to the terrestrial trophic web of the Malpelo Fauna and Flora Sanctuary, Colombia malpilensis, the dotted galliwasp D. millepunctatus, the ant Odontomachus sp., and the Isopoda reflect high δ 13 C signals in their tissue from detritus (Fig. 5, Table 2). The δ 13 C present in terrestrial C 3 plants is mostly integrated and synthesized by the Orders Hymenoptera, Diplopoda, and Microcoryphia, which showed low δ 13 C values reflecting assimilation of detritus (Fig. 5, Table 2).
The organic matter present in the detr itus (δ 13 C = − 20.1 to − 17.3‰) reflected the isotopic values of organic matter of marine origin (Fig. 6). The basal δ 13 C detritus of the terrestrial ecosystem showed no statistical difference compared to phytoplankton δ 13 C values (δ 13 C no corrected ; Wilcoxon rank sum test, W = 20, P = 0.80), macroalgae (δ 13 C no corrected ; Wilcoxon rank sum test, W = 13, P = 0.16), crustaceans (δ 13 C Corrected*TDF ; Wilcoxon rank sum test, W = 52, P = 0.29), and S. granti eggs (δ 13 C Corrected*TDF ; Wilcoxon Rank sum test, W = 19.5, P = 1). In contrast, the basal δ 13 C detritus and δ 13 C of Fig. 5 A A biplot of the sources (colored diamonds [mean ± SD]) and consumer (grey points) was used to run the stable isotope mixing models. B Estimation of the contribution probability (in %) of the terrestrial basal sources to the diet of terrestrial secondary consumers.
C Estimation of the contribution probability (in %) of the terrestrial basal sources to the terrestrial ecosystem in Malpelo Fauna and Flora Sanctuary, Colombia zooplankton were statistically different (δ 13 C Corrected*TDF ; Wilcoxon rank sum test, W = 109, P = 0.002) (Fig. 6).

N-enrichment
Enrichment analysis indicated that terrestrial C 3 plants, millipedes, crickets, and Hymenoptera had high degrees of depleted 15 N relative to the 15 N of eggs and feathers of S. granti. However, millipedes, crickets, and Hymenoptera reflect 15 N-enriched tissues relative to detritus. Conversely, the rest of the terrestrial components of the Malpelo FFS are enriched in 15 N relative to the reference components, i.e., detritus, and eggs and feathers of S. granti (Fig. 7).

Isotopic niche and isotopic overlap
The wide isotopic range of carbon, and especially of nitrogen, in the terrestrial ecosystem reflected an isospace with a total area (TA terrestrial ) of 134.7‰ 2 and an isotopic niche (SEA C_terrestrial ) of 30.4‰ 2 (Fig. 8A). After excluding terrestrial C 3 plants, the isospace and the isotopic niche were 65.1‰ 2 and 17.3‰ 2 , respectively (Fig. 8B). The isospace and isotopic niche of the marine ecosystem (TA marine = 117.2‰ 2 and SEA C_marine = 21.0‰ 2 ) were very similar to those of the terrestrial ecosystem, excluding C 3 plants (Fig. 8A, B).
Considering the low contribution of terrestrial C 3 plants to the terrestrial trophic web, two isotopic overlap scenarios were considered: one including C 3 plants and one excluding them. The terrestrial (C 3 plants; red box, Fig. 8A) and marine isospaces reflected an isotopic overlap of 0.85 (SIBER overlap; Fig. 8A), suggesting an overlap probability of 65% (nicheROVER) between the two ecosystems. In contrast, the marine isospace indicated a higher overlap probability with the terrestrial isospace (76%; Fig. 8A). In the second scenario, the estimated isotopic overlap between the terrestrial and marine isospaces was 0.71 (SIBER overlap; Fig. 8B), corresponding to 82% (terrestrial vs. marine) and 70% (marine vs. terrestrial) overlap between the two isospaces (Fig. 8B).

Discussion
The allochthonous organic matter produced in an ecosystem and transferred into another plays an important role in food webs and ecosystem dynamics (Polis et al. 1997a). The food web and the ecological dynamics of the terrestrial ecosystem of the Malpelo SFF are modulated by detritus. This suggestion is supported by the similarity of the δ 13 C Terrestrial with the marine basal sources (Fig. 6) and the enrichment in 15 N of the terrestrial primary and secondary consumers (Fig. 7). By integrating isotopic values with a marine origin, terrestrial consumers reflect a high similarity in the terrestrial and marine isospaces i.e., isotopic overlap > 70%. This study demonstrated that the terrestrial food web is greatly supported by detritus, suggesting that the terrestrial ecosystem presents a high interaction and dependence on the marine environment.
Terrestrial macro-species (i.e., A. agassizi, D. millepunctatus, and J. malpilensis) had similar δ 13 C values to those of the marine ecosystem. This is the result of the high trophic interaction between these species and the seabird S. granti and its derivates; this seabird is responsible for transporting and depositing energy and matter of marine origin on terrestrial ecosystems (López-Victoria and Werding 2008;López-Victoria et al. 2009, as has been observed on islands in the Gulf of California, Mexico (Anderson and Polis 1999;Sánchez-Piñero and Polis 2000), in Baccalieu Island, Canada (Duda et al. 2020), the Pacific and Indian Oceans, as well as in the Mediterranean Sea (Caut et al. 2012). Such a marked link between the ecosystems (terrestrial and marine) of the Malpelo FFS could be due to (1) a high similarity between terrestrial (excluding terrestrial C 3 plants; Fig. 8B) and marine isospaces, which reflects a high trophic interaction (high isotopic overlap) between both ecosystems. This can be explained by the high similarity between the δ 13 C of terrestrial detritus and those of the marine primary producers (i.e., macroalgae: − 21.0 to − 16.0‰; phytoplankton: − 20.7 to − 15.5‰ [this study]), as well as S. granti eggs, marine macroalgae, and marine crustaceans from the Malpelo FFS as a result of the transport and deposition of nutrients by S. granti (López-Victoria and Werding 2008;López-Victoria et al. 2009. These are harnessed by the majority of terrestrial consumers, which is supported by the high contribution of detritus to terrestrial consumers (Fig. 5B) and the low input of terrestrial primary producers (i.e., terrestrial C 3 plants) as a result of their low coverage of the island (< 0.02 km 2 ; López-Victoria and Rozo 2006). This supported the food chains of small organisms that mostly consume vegetable matter i.e., the orders Diplopoda (millipedes) and Microcoryphia ( Fig. 5B; Bueno-Villegas 2012; Bach de Roca et al. 2015); and 2) differences in the enrichment values of 15 N in the various terrestrial components, such as A. agassizi, D. millepunctatus, J. malpilensis, Isopoda, Araneae, Lumbricullidae, and Odontomachus sp., which were 15 N-enriched concerning detritus and S. granti's eggs and feathers (Fig. 7). This could be the result of two factors: (i) The decomposition of naturally 15 N-enriched guano and seabird tissue (Anderson and Polis 1999) could be further 15 N-enriched due to the volatilization of 14 N (Lindeboom 1984;Mulder et al. 2011) and to the fast mineralization of uric acid to ammonium (NH 4 + ) from guano (Wainright et al. 1998). This would lead to a greater isotopic fractionation, provoking 15 N-enrichment of the residual NH 4 + reservoir (Mizutani and Wada 1988;Wainright et al. 1998); and (ii) high δ 15 N values of these terrestrial animals are related to the elevated consumption of prey from high trophic levels, as is the case of S. granti, which occupies high trophic positions in the marine environment (3.9; Cortés 1999). Therefore, primary (i.e., Isopoda and ants Odontomachus sp.) and secondary consumers (i.e., A. agassizi, D. millepunctatus, and J. malpilensis) incorporate 15 N directly from the consumption of S. granti and its byproducts (López-Victoria and Werding 2008;López-Victoria et al. 2009. The terrestrial C 3 plants, crickets, millipedes, and Hymenoptera were 15 N-depleted regarding detritus and S. granti's eggs and feathers. These results suggest that these components have values of δ 15 N compatible with terrestrial primary production (Craine et al. 2009;Amundson et al. 2003), which is consistent with their food preferences (Bueno-Villegas 2012; Bach de Roca et al. 2015). Plants fertilized with guano have 15 N-enriched values (Anderson and Polis 1999) as is the case of the islands in the Gulf of California (C 3 plants = 24.5 ± 1.1 ‰, C 4 = 24.3 ± 1.4‰; Barrett et al. 2005). The terrestrial C 3 plants of Malpelo FFS showed low δ 15 N values (7.4 ± 2.25‰). Values found for these terrestrial C 3 plants of the Malpelo FFS are consistent with atmospheric nitrogen fixation and were depleted in 15 N relative to the eggs and feathers of S. granti (Fig. 7). Similar results were reported for Possession Island in the Indian Ocean (mean ± SD: plants = 5.2 ± 1.05‰, seabirds = 9.3 ± 0.45‰, enrichment = − 0.44; Caut et al. 2012). This suggests that terrestrial C 3 plants of Malpelo FFS obtain N directly from the atmosphere.
The above suggests that the S. granti colony positively impacts terrestrial communities of Malpelo FFS due to the high contribution of guano and other "byproducts" that terrestrial species consume directly. This is reflected in the high abundance of J. malpilensis (estimated population: 833,000 individuals; López-Victoria and Werding 2008), D. millepunctatus (12,000-18,000 individuals; López-Victoria et al. 2011), and A. agassizi (60,000-102,000 individuals; López-Victoria et al. 2011) in the island. In contrast, the large S. granti colony could negatively affect the population of terrestrial plants by reducing their cover on the island. This phenomenon has been observed on Malpelo island (S. Bessudo Lion, personal communication) and could be related to: (1) the high concentrations of guano during the dry season that could exceed the concentration limits of essential nutrients and eventually toxify the soil and limit the development of plants; this could also prevent the establishment of native plants in places where there is a high density of seabirds (Boutin et al. 2011;Sánchez-Piñero and Polis 2000) and (2) the reduction of nutrients due to guano washing off during the rainy season, which limits soil formation and affects the adequate development of plants (Caita and Guerrero 2000).
In conclusion, the δ 13 C and δ 15 N values in terrestrial organisms from Malpelo FFS originate mainly from the marine environment confirming the close link and high trophic interaction between the marine and terrestrial ecosystems of Malpelo FFS as suggested in previous studies (Wolda 1975;López-Victoria et al. 2009). The values of terrestrial components are associated with food chains supported mainly by organic matter decomposition processes (i.e., consumption of detritus) with low compatibility with a primary productivity-based diet (i.e., C 3 plants).
The transport of nutrients from sea to land in Malpelo FFS is governed mainly by S. granti. This suggests that the terrestrial ecosystem dynamics of the island are largely influenced by an important control generated by the "donor" habitat (marine ecosystem) over the "receptor" habitat (terrestrial habitat) due to the transport and contribution of matter and energy between ecosystems (Polis et al. 1997a). However, there may be two additional pathways of nutrient inputs at the sea-land interface of the Malpelo FFS such as the consumption of marine algae and marine crabs (Grapsus grapsus) by J. malpilensis and D. millepunctatus through the intertidal zone (López-Victoria et al. 2009, as well as terrestrial nutrients (e.g., organic matter, seabird guano, etc.) transported to sea due to runoff during the rainy season between May and December (annual precipitation ~ 2500 mm; von Prahl 1990; López-Victoria and Estela 2007).
Terrestrial nutrients could affect primary producers locally, altering the typical values of marine primary productivity surrounding the Malpelo FFS and modifying seasonal marine trophic dynamics (Ishida 1996;Wait et al. 2005). As a result, this would be reflected in their isotopic values. Despite the contributions of terrestrial nutrients to the sea and the effects that these contributions may have on the dynamics of this ecosystem, more studies are necessary to validate these hypotheses and identify other trophic connectivity routes between the terrestrial and marine ecosystems of Malpelo FFS. Nevertheless, these sources of marine nutrient inputs into the terrestrial ecosystem have not been studied in detail. More studies are necessary to estimate the intertidal zone and terrestrial ecosystem's contribution to Malpelo FFS. In turn, this would improve ecological knowledge regarding the dynamics of this small oceanic island.
Finally, given the impact exerted by the donor habitat on the receptor habitat, an eventual disturbance of marine populations may alter food webs due to the transitional interphase between the marine and insular environment (Sullivan and Manning 2019). The present study documented trophic interactions between marine and terrestrial ecosystems, supporting how diverse species can cross the limits of distinct environments (e.g., terrestrial and aquatic). Furthermore, this study showed how SIA constitutes a useful tool in the identification of trophic interactions between terrestrial and marine ecosystems.