Invasive species in Weddell Sea : Effects on food web structure

Linköping University | Department of Physics, Chemistry and Biology Type of thesis, 60 hp | Educational Program: Physics, Chemistry and Biology Spring term 2018 | LITH-IFM-A-EX—18/3468--SE

Invasive species in Weddell Sea

Effects on food web structure

Inger-Marie Wohlfarth

Examinator, Karl-Olof Bergman Tutor, Anna Eklöf

Datum Date

2020-04-20 Avdelning, institution

Division, Department

Department of Physics, Chemistry and Biology Linköping University

URL för elektronisk version

ISBN

ISRN: LITH-IFM-A-EX-18/3468-SE

_________________________________________________________________ Serietitel och serienummer ISSN

Title of series, numbering ______________________________

Språk Language Svenska/Swedish Engelska/English ________________ Rapporttyp Report category Licentiatavhandling Examensarbete C-uppsats D-uppsats Övrig rapport _____________ Titel Title

Invasive species in Weddell Sea – Effects on food web structure

Författare

Author

Inger-Marie Wohlfarth

Nyckelord

Keyword

Centrality, Climate change, Food webs, Google PageRank, Invasive species, Structure, Weddell Sea

Sammanfattning

Abstract

The cold water of Antarctica has a unique endemic fauna, where durophagous predators are rare or absent. Due to climate change the water is heating up and the predators have begun to return to the Southern Ocean, which could bring a lot of changes to the food web. There is a high risk it will lead to losses in the unique marine fauna of Antarctica. The aim of this study is therefore to examine the potential effect these invasive species has on the food web structure in the Weddell Sea. To study this, several general network metrics were used (connectance, number of interactions, vulnerability and generality, trait distributions), as well as a number of centrality metrics (betweenness, closeness, PageRank). The analyses showed that none of the invasive species become important in the Weddell Sea food web. Nor do they significantly change the food web structure in any way which impact the importance of the native species. Their great opportunism regarding their prey species, and thereby their connectedness and thus their position in the network, are probably the main reason why theses invasive species did not become important in this food web. The lack of changes in the food web structure due to the presents of these invasive species are probably also a result of not including factors such as abundances and network dynamics in the analyses, which seem to be the driving forces when it comes to changes in food web structure caused by invasion of species.

Contents

3.3 General network metrics . . . 7

3.4 Centrality metrics . . . 8

4 Results 11 4.1 General network metrics . . . 11

4.2 Centrality metrics . . . 16

5 Discussion 21

6 Conclusions 26

7 Acknowledgements 27

1

Abstract

The cold water of Antarctica has a unique endemic fauna, where durophagous predators are rare or absent. Due to climate change the water is heating up and the predators have begun to return to the Southern Ocean, which could bring a lot of changes to the food web. There is a high risk it will lead to losses in the unique marine fauna of Antarctica. The aim of this study is therefore to examine the potential effect these invasive species has on the food web structure in the Weddell Sea. To study this, several general network metrics were used (connectance, number of interactions, vulnerability and generality, trait distributions), as well as a number of centrality metrics (betweenness, closeness, PageRank). The analyses showed that none of the invasive species become important in the Weddell Sea food web. Nor do they significantly change the food web structure in any way which impact the importance of the native species. Their great opportunism regarding their prey species, and thereby their connectedness and thus their position in the network, are prob-ably the main reason why theses invasive species did not become important in this food web. The lack of changes in the food web structure due to the presence of these invasive species are probably also a result of not includ-ing factors such as abundances and network dynamics in the analyses, which seem to be the driving forces when it comes to changes in food web structure caused by invasion of species.

Keywords: Centrality, Climate change, Food webs, Google PageRank, Invasive species, Structure, Weddell Sea

2

Introduction

During the last 50 years the impact of invasive species has been an increas-ing concern (Gurevitch & Padilla 2004). Studies have shown that biological invasions can affect whole ecosystems in diverse ways, both positively and negatively, depending on different variables (Gallardo et al. 2016, Jeschke et al. 2014). Invasive species interact with native species in the ecosystem they have invaded and create new trophic links, which can affect the system due to possible changes in abundance and demography of the native species (David et al. 2017). The invasive species do not only affect local species in a food web but can also alter the food webs structure due to the effects of the invasion and changes can propagate through the whole web (David et al. 2017).

All species on Earth are parts of different ecosystems, where they interact with each other and the environment around them (Newman 2010). Organ-isms who prey on each other in a given ecosystem form a web of interactions between species, e.g. a food web (Newman 2010). Food webs are a type of ecological network where the species in the ecosystem are represented as nodes, which are connected to each other representing their feeding interac-tions. A directed edge from node A to B indicates that species A consumes species B. This relationship can also be drawn the other way around, showing how energy flows through the food web (Newman 2003). To usefully repre-sent the relationship between species in a food web a matrix is often used. The simplest way to construct these matrices is for columns to represent prey or resources and rows to represent predators or consumers. If species i feeds upon species j, a 1 is assigned to the cell at row i and column j, and if species idoes not feed on species j a 0 is assigned to the cell. The usage of a matrix can be at aid for quantitative analyses of food web structure when working with interactions between species (Pascual & Dunne 2006)

To describe these food webs, basic structural metrics are often used. One of them is connectance, the number of observed trophic links in a food web di-vided by the total number of possible links (Murtaugh & Derryberry 1998). In a more technical sense this refers to the percentage of non-zero elements in an interaction matrix (Gardner & Ashby 1970). Two other basic structural

metrics are generality and vulnerability, where generality is the number of prey species a species has, and vulnerability is the number of predator species a species has (Schoener 1989). There are also other metrics used when study-ing food web structure. Centrality metrics look at the nodes or edges in the network and quantify how important they are in the network regarding to their position to other nodes and edges (Newman 2010). This study uses three dif-ferent types of centrality measures to look at the structure, betweenness cen-trality, closeness centrality and importance/PageRank cencen-trality, besides the basic metrics described above. Betweenness centrality (xi) is a metrics which

measure the number of geodesics (shortest paths) from one node in the net-work to another, going through a third node. A node has high betweenness if it lies on the shortest path between many pairs of nodes (Allesina & Pascual 2009). The metrics closeness centrality (C_i) is defined as the inverse of the mean distance from one node to other nodes in the network (Newman 2010). Nodes are highly central (close to many other nodes) if they have short dis-tances (paths) to many nodes in the network (Allesina & Pascual 2009). The PageRank centrality, which is a modified version of the Google PageRank al-gorithm, rank species as important if they directly or indirectly consume other species that in turn also are important Allesina & Pascual (2009).

Food webs provide complex yet manageable illustrations of biodiversity, in-teractions between species and ecosystem structure and functions (May 1986, Levin 1992). Research on ecological networks is an important and rapidly ad-vancing field, both with regard to the response of disturbances to ecosystems including species invasions and biodiversity loss (Pascual & Dunne 2006), and also as describing the constrains or processes originating general struc-tural patterns (Brose et al. 2004). Ecological networks are useful tools for studying impacts of invasive species, since they are simplifications of the ecol-ogy of systems but still can capture information necessary to understand and predict impacts of invasive species (David et al. 2017).

Definitions of invasive species today differ a lot and there is no single def-inition that is widely accepted. This is due to the variance in attribute and consequences for the invaded ecosystem of the invader (Pantel et al. 2017). The definitions have different combinations of criteria based on impact, ori-gin and demography (Blackburn et al. 2011, Gurevitch et al. 2011). David et al. (2017) adopted following definition for invasive species: a nonnative

species that is introduced, able to increase in population size, spread in space and maintain in the introduces area, which also is the definition used in this study. Invasion of food webs have the potential to alter the web by causing severe changes to its biodiversity (Salvaterra et al. 2013). Direct predation is one of the effects of invasion that affects ecosystem the heaviest, invasive species can decrease populations of native species which they prey upon and even cause their extinction (Bruno et al. 2005, David et al. 2017). Declines in indigenous species can also come from competition for food and appar-ent competition with invasive species, however, this rarely lead to extinctions (David et al. 2017). Loss of a single species in an ecosystem can lead to other problem than the loss itself, as it can lead to a cascade of secondary extinc-tions which can change the trophic structure of the ecosystem (Paine 1966, Estes & Palmisano 1974, Borrvall et al. 2000, Sole & Montoya 2001, Dunne et al. 2002, Ebenman et al. 2004, Koh et al. 2004). The largest number of secondary extinctions are triggered by the loss of species at low trophic levels and species with high connectivity (species with many links to other species in the web), and increasing connectance in the food web decrease the number of secondary extinctions (Eklöf & Ebenman 2006). Exactly how these losses affect the ecosystems will depend on the complexity of the disturbed ecosys-tem, the number of species lost, as well as their specific functions (Dunne et al. 2002). Knowledge about invasion of species will provide solutions of fundamental evolutionary and ecological problems, as well as contribute to control, prevent and predict invasions more effective and efficient (Gurevitch et al. 2011).

In this study I specifically analyses the potential impact of invasive species in the Weddell Sea, which is a part of the Southern Ocean. The Southern Ocean is a very special ecosystem that surrounds Antarctica. Its unique food web structure developed during the late Eocene, about 40 million years ago, during a global cooling. Due to this cooling most of all fast-moving skeleton-crushing bony fish, sharks and crabs (durophagous predators) were excluded, which left an endemic living fauna of slow-moving invertebrates as top preda-tors and epifaunal suspension feeders dominating many of the soft-substratum communities. The cooling created a physical barrier that isolated the Southern Ocean to biological invasion and prevented most of the durophagous preda-tors passing through, due to limitation in their mobility because of the cold water (Aronson et al. 2007).

Over the past 140 years the average temperature of Earth’s climate has in-creased by approximately 0.85 degrees Celcius (Intergovernmental Panel on Climate Change 2014), and changes in global climate have led to changes in species geographic ranges, behaviour and phenology (Helmuth et al. 2006, Smetacek & Nicol 2005, Walther et al. 2002, FIelds et al. 1993, Clarke et al. 2007). Introduction and establishment of invasive and exotic species are ex-pected to increase due to higher temperatures (Stachowicz et al. 2002). To-day’s climate change is making the physical barrier surrounding the Southern Ocean disappear, which has led to invasions of nonative species into parts of Antarctica (Aronson et al. 2007). During recent years different species of durophagous crabs have arrived to the Southern Ocean and and accomplished to invade parts of it’s ecosystem (Ahyong & Dawson (2006),Thatje & Lörz (2005), Thatje (2005)). As the sea temperatures continue to rise there is a high probability that more durophagous predators will return to the South-ern Ocean as well. The return of these predators can potentially bring large changes to the ecological system, and there is a high risk that the unique en-demic fauna in Antarctica will be lost (Aronson et al. 2007). The aim of this study is therefore to examine how these invasive species affect the food web structure in the Weddell Sea. Where focus lies on how the invasive species af-fect the food web structure connected to centrality, but also how the structure is affected associated with general metrics.

3

Methods

3.1

Data

The Weddell Sea, a part of the Southern Ocean, is located between 74-78 degrees south between the Antarctic Peninsula in the west and the Coats Land of East Antarctica in the east. It has a length of about 450 kilometers and a water depth that varies from 200 meters to 500 meters. Shallower parts in the eastern and south part of the Weddell Sea, which form the coastline, are covered by continental ice that lower the shelf edge to 500 - 600 meters below sea level.

Data from the Weddell Sea has been collected since 1983 and is one of the most mapped marine food webs, although seasonal changes have been ig-nored. The food web contain 488 species and 16041 feeding interactions. Each species was observed from a combination of field studies between year 2001 and 2004, where stomach content was analysed to learn their diet com-position (Jacob 2005). To study the impact of invasive species to this food web, 12 species were added to the existing web. Six crab species and six shark species that were thought to be able to invade this ecosystem (U. Jacob personal communication). Their feeding interactions were estimated though knowledge of which species they interact with in other food webs.

The data set consists of different trait data connected to each species as well as data of how all species interact with each other. Six traits were included, both categorical and continuous; body size, metabolic category, feeding type, feed-ing mode, mobility type and (feedfeed-ing) environment. The different categories within the traits can be found in Table 1.

Table 1: The trait of the species in the Weddell Sea.

Body size Categorised by their weight in grams.

Metabolic category Primary producers, invertebrates, endotherm vertebrate and ectotherm vertebrate.

Feeding mode Deposit-feeder, grazer, predator, predator/scavenger, pri-mary producer, suspension-feeder.

Feeding type Herbivorous/detrivorous, carnivorous, carnivo-rous/necrovorous, omnivorous, detrivorous or primary producer.

Mobility type Sessile, swimming, walking.

Feeding environment Bathydemeral, benthic, demersal, pelagic, benthopelagic, land-based.

3.2

Data preparation

All work with the data and all analyses has been done within the program R, version 3.3.1 (R Core Team 2016). The data were divided into two different types of files. One containing edge lists with two columns representing the interactions between species in the food web, where column 1 represent the consumers and column 2 the resources. The second file contained all the species traits, with species as rows and their traits as columns. All data was converted into two food webs, one including the invasive species and one without.

3.3

General network metrics

First, I performed a number of general network analyses, described below, to get an overview of the food webs, and also to see if the two data set differed from each other. From the edge lists, adjacency matrices Ai jwere made with S

rows and S columns, representing the links in the food web with S species. A 1 was assigned to Ai j if species i (row) consumed species j (column), otherwise

a 0 was assigned. The number of non-zero elements in the adjacency matrix represented the number of feeding interactions L in the food web.

The general measurement I started with was connectance (c) of the networks, proportion of potential trophic links, which was calculated through

c= L/S2. (1)

Then the trophic level (T ) for each species in the food webs was calculated. This equation was used to calculate the species trophic level

T = v(I − A)−1, (2) where A is a normalized adjacency matrix of a acyclic food web. It is normal-ized so the sum of all rows would be equal to 1. I is an identity matrix with S rows and A columns and v was a vector of ones with the length of S.

Generality and vulnerability were calculated for each food webs as well. Gen-erality (G) is defined as the mean number of prey per predator and vulnera-bility (V ) is defined as the mean number of predator per prey.

3.4

Centrality metrics

Three types of centrality metrics were used to learn more about the food webs structural features; betweenness, closeness and eigenvector (PageRank) cen-trality.

Betweenness and closeness centrality

Betweenness centrality (xi) is the number of geodesics from one node to

an-other going through a third node, and was calculated through following equa-tion

x_i=

_∑

st

ni_st

g_st. (3)

The variable ni_stis the number of geodesic paths from s to t that passed through i, and g_st is the total number of geodesic paths from s to t (Newman 2010). Closeness centrality (C_i) is the inverse of the mean distance from one node to other nodes, and was calculated through following equation

C_i= n ∑jdi j

. (4)

The variable n is the number of nodes in the network and d_{i j} is the the length of a geodesic path from i to j (the number of edges along the path) which then is averaged over all columns ( j) in the network (Newman 2010).

PageRank centrality

I also used a centrality measure that is used to rank the importance of web pages in Google search, (Google) PageRankT M. The algorithm rate pages important if other pages that in turn also are important points at them. Since an ecological network, which I study, is not the same as the network of pages in the World Wide Web, a modified version of the algorithm is used to work for ecological networks (Allesina & Pascual 2009). The modified algorithm

of PageRank by Allesina & Pascual (2009), rank a species as important if it is directly or indirectly connected to other species that in turn also are important. It is similar to the ranking of web pages, but the importance moves in the opposite direction, species are important if they point to important species (get consumed by important species).

One problem with applying this algorithm on food webs, are that they nei-ther are irreducible nor primitive which are criteria for the algorithm to work (Allesina & Pascual 2009). However, there exist simple biological solutions for these problems. All matter in the food web are channelled through feeding pathways through the web, originated from primary producer who receive it from the surrounding environment. And every species in the food web also has an intrinsic loss of matter, which represent the build-up of detritus that partly is recycled into the food web (Allesina & Pascual 2009). Through applying this biological knowledge to the algorithm any food web becomes primitive and irreducible. It is applied through attaching the network to a special node, a ”root”, that is connected to all primary producers (they eats the root) and then also add a link from each node the root (the root eats all species) (Allesina & Pascual 2009). When the algorithm is modified this way, it is possible to assign importance by calculate the eigenvector (v) associated with the dominant eigenvalue (λ∗= 1) (Allesina & Pascual 2009).

The PageRank centrality is calculated from the edge list, according the mod-ified algorithm (Allesina & Pascual 2009). In my data the taxon ”Sediment” represent the buildup of detritus and become the root node and connected as consumer to all species in the food web, as well as the resource of all basal species. From the modified edge list an adjacency matrix was created, nor-malized by column, and then the leading eigenvector was obtained from the matrix and normalized so the sum of all entries would be equal to 1.

Sensitivity analysis of PageRank centrality

The feeding interactions connected to the invasive species were estimated into the food web and therefore there existed a risk that the result of the PageRank centrality were affected. I was interested in how sensitive the result was to

changes in the network structure. To examine how sensitive the result from the PageRank centrality was, several sensitivity analyses were performed.

From the original edge lists; 5 %, 10 % and 15 % of the interactions were randomly removed and new PageRank values were produced from the new edge lists. This was repeated 1000 times for each percentage.

If the removal of the random interactions from the original edge lists affected the food web in the way that it split into two food webs, the random removal was repeated. The removal was repeated until a fully connected food web without sub-graphs was created before new PageRank values were produced.

4

Results

4.1

General network metrics

The two data sets, the one without the invasive species and the one with do not differ a lot comparing the results from the general network analyses (Table 2).

Table 2: Results from general network metrics of the food web of the Weddell Sea, both including estimated invasive species and without.

Without invasive sp.

With invasive sp. Number of directed links/edges (L) 16041 17068

Number of species/nodes (S) 490 502

Connectance (c) 0.0668 0.0677

Mean / Median number of links a

species has 65.5 / 56 68 / 60

Mean / Median number of prey links a

species has 33 / 7.5 34 / 9.5

Mean / Median number of predatory

links a species has 33 / 26 34 / 31

We looked at the position of the invasive species in the different trait dis-tributions to see if they distinguished from the native species. The invasive species were found in certain areas and categories of the traits, but their traits and combination of them were not unique compared with the native species (Figure 1 to 3). Looking at the interactions, the invasive species all had large numbers of prey species and low or non-existing number of predators (Table 3). Their number of feeding interactions and prey species were for a

major-ity of the invasive species above the average numbers, and their number of predators were below (Table 2).

Table 3: The invasive species that is thought to invade the Weddell Sea food web and in this study was added to the food web, and their number of prey and predator species. The first six represent the crab species and the rest represent the shark species.

No. of prey No. of predators Paralomis granulosa 99 3 Paralomis birsteini 99 4 Halicurcinus plantus 10 4 Hyas araneus 74 3 Lithodes murrayi 117 3 Neolithodes diomedeae 117 3 Somniosus antarcticus 80 0 Centroscymnus coelolepis 102 0 Lamna nasus 56 0 Etmopterus granulosus 94 0 Squalus acanthias 185 0 Carcharodon carcharias 89 0

The distribution of the species logged body weight showed a high frequencies around zero and then it rapidly decreases on both sides (Figure 1). The inva-sive species are identified among the heavier organisms in the distribution.

Figure 1: Distribution of the species in the Weddell Sea food web according to their body weight, where A) shows the food web without estimated invasive species and B) the food web with. The body weights were transformed by the natural logarithm to more easily display the distribution, due to that the food web contains very large and very small organisms. The red marking in the distribution represent the area where the body weight of the invasive crabs is located, and the blue marking represent where the body weight of the invasive sharks is located.

Looking at the categorical traits the invasive species were located as ectotherm vertebrates (the invasive shark species) and invertebrates (the invasive crab species) when it came to their metabolic category. All invasive species had carnivore as their feeding type, and they were identified as predators (the sharks) and predators/scavengers (the crabs) as their feeding mode (Figure 2).

Figure 2: Distribution of the species in the Weddell Sea food web according to these traits; Metabolic category, Feeding type and Feeding mode. The dark grey parts of the bars represent the native species in the web, and the light grey represent the estimated invasive species.

In the distribution of the different movement types the invasive species were located as swimming (the sharks) and walking (the crabs). For the feeding environment they were located to be in the benthic, benthopelagic and pelagic environment to feed (Figure 3).

Figure 3: Distribution of the species in the Weddell Sea food web according to these traits; Movement type and (Feeding) Environment of the species in the Weddell Sea food web. The dark grey parts of the bars represent the native species in the web, and the light grey represent the estimated invasive species.

4.2

Centrality metrics

The result from the closeness analysis without adding the invasive species to the food web showed that all the species had quite similar centrality values. There were only small shifts that determined the ranking of the species, except for one specie in the top, Orcinus orca (Figure 4). Due to this small or non-exiting change in the ranking values, several species had the same rank of closeness centrality. When the invasive species were added to the food web, all except one of them ended up in the top of the ranking scale. However, the small or non-existing changes in ranking values did not change a lot when the invasive species were added, which makes it difficult to determine the actual scores of all species since several have the same values or just very small changes compared with the one above or below.

Figure 4: The 25 species with highest closeness centrality in the Weddell Sea food web. The plot to the left with dark grey bars represent the ranking without the invasive species, and the plot to the right with light grey bars represent the ranking with the invasive species included in the food web. The species names with an asterisk are the invasive species.

The result from the betweenness centrality did not show any significant dif-ferences between the food web including the invasive species and the one without them (Figure 5). All invasive species got a centrality value of zero and their presences didn’t change the ranking of the other species in the food web in any notable way.

Figure 5: The 25 species with highest betweenness centrality in the Weddell Sea food web, where the invasive species were included. The dark grey bars represent the centrality values that the native species have when the invasive species are not included, and the white bars overlapping the grey represent the centrality values when the invasive species are included in the food web.

Both the PageRank analysis of the food web including the invasive species and the one without them had small differences in the PageRank values be-tween species in the middle of the ranking, that increases toward the top and the bottom of the ranking. Comparing the two PageRank analyses from the food web with the invasive species and the one without them showed only small changes in the PageRank values. All PageRank values decreased slightly when the invasive species were added to the food web, but the ranking order still remained almost intact. All invasive species that were added to the food web ended up among the 65 lowest ranked species, where all the shark species had the lowest PageRank values off all species.

Three sensitivity analyses was performed to make sure the results from the PageRank analysis of the food web where the invasive species were included, did not depend on the estimation of the invasive species and their estimated interactions. The results showed that the result from the PageRank analysis were stable and not affected by the estimation of the new feeding interactions (Figure 7). They showed similar results of the species ranking compared with the original result where no species interaction had been randomly removed. The result from the original PageRank analysis overlap with the result from the different sensitivity analyses where the frequency were at its highest (Fig-ure 7). Results from the sensitivity analyses where 5 % and 15 % of the interaction had been removed can be found in the appendix (Figures 8 and 9).

Figure 6: The 25 species with highest PageRank centrality in the Weddell Sea food web, where the invasive species were included. The dark grey bars represent the centrality values that the native species have when the invasive species are not included, and the white bars that overlap the grey bars represent the centrality values when the invasive species are included in the food web. The species names with an asterisk are the invasive species.

Figure 7: Sensitivity analysis for the Weddell Sea food web including the invasive species, where 10 % of the interactions were randomly removed. The black dots represent the original PageRank value, when no interactions were removed. The figure shows the 19 species with the originally highest ranked PageRank values.

5

Discussion

The impact of invasive species can affect ecosystem in different ways. One well-known effect is extinctions, mostly through direct predation (David et al. 2017). An example of this is the extinctions of haplochromine species in Lake Victoria in 1970s due to predation by adult Nile Perches which invaded the ecosystem (Witte et al. 1992, Goudswaard et al. 2008). Extinctions due to invasive species can also be caused by other causes than direct predation. Trophic causes like competition for resources and apparent competition are also impacts that can cause extinctions in an ecological network, as well as combinations and interplay of these different trophic impacts (David et al. 2017). If the shark and crab species manage to invade the Weddell Sea food web, they will to some extent impact the food web through direct predation, competition of food sources and indirect competitions, and this impact have the possibility to propagate through the food web and affect a number of species in different ways (David et al. 2017).

In this study, however, none of the invasive species becomes important in the Weddell Sea food web based on the PageRank centrality. Nor do they signifi-cantly change the food web structure in the perspective of the centrality mea-surements used here. The results from the general network metrics showed similar patterns with lack of significant differences due to the addition of these invasive species. There were only small increases in the values where the in-vasive species were included compared with those without them, but not in any considerable way. This was however expected due to the size and com-plexity of the food web and since only twelve species were added. Neither in the distribution of the species traits could any notable differences be dis-tinguish due to the presence of these invasive species. The invasive species were limited to certain areas within the trait distributions, but did not have any unique traits or combinations of traits compared with the native species.

Research until now have focused on looking at the local effects invasive species has on food webs, and few studies has studied the impact they have on the whole food web structure (David et al. 2017). By using centrality metrics, I had the possibility to get a better understanding of how central the native species were in the food web structure compared with the invasive species,

and how this would change with the positions of the invasive species. Could their invasion and the positions they end up in impact the food web further than the species closest connected to them? As stated in the paragraph above the result from the PageRank centrality as well as the betweenness centrality showed that the invasive species do not take important positions in the struc-ture of the Weddell Sea food web. Nor do they change the food web strucstruc-ture in any way that impact the ranking order and thereby the importance of the native species. They them self ended up in the bottom of the ranking order in both the PageRank and the betweenness centrality. In other words, the in-vasive species will not become keystone species in the food web or a bridge where a large amount of energy flow through. Their low PageRank scores in-dicate that they are not important in the perspective of robustness. Thus they do not take a position in the structure that is important for the robustness of the food web. In a position when removed can lower the food web robustness and cause a series of secondary extinctions. As well as a position which have the possibility to affect surrounding species in the food web, which in turn also are important for the food web robustness.

Neither in the analysis of the closeness centrality do the invasive species change the food web structure in any way that impact the ranking order, other than taking a place in it. Majority of the invasive species do, however, end up in the top of the ranking order for this centrality metrics. This suggest that they are close to the other species in the food web through all interaction paths. Which in turn can imply that their impact can reach broadly over the food web and propagate further on. However, the local impact the invasive species have on the species closest to them and the effects that can come of their impact has not been included in this study. But since they are closely connected through their feeding interaction with the most other species in the food web, their impact has the possibility to propagate through large parts of the food web and they are likely to give some kind of effect on the network dynamics. Potential effects for the Weddell Sea community could be changes in interaction strength and abundance, through changes in diets and even trig-gered extinctions if they impact one or more of the keystone species in the food web. How large and comprehensive the effects will be, as well as ex-actly how the food web will react can only be speculated in. It is, however, clear that the invasive species have a potential to impact the food web from

their predicted positions in the food web, but they do not change the structure and positions of the important species in the web.

Currently there exist very little information of the diet of the deep water litho-dids living south of the polar front, and their feeding behaviour (Griffiths et al. 2013). Recent studies claim that invasion of these durophagous (bone crushing) lithodid crabs threatens the Antarctic shallow marine ecosystems, and their native fauna of weakly calcified benthic invertebrate communities (Thatje et al. 2005, Aronson et al. 2011, Smith et al. 2012). However, it re-mains uncertain if the invasion will increase the risk of predation on sedentary invertebrates and through that also decrease the native fauna of these ecosys-tems (Griffiths et al. 2013). Many of the regions deep water lithodids are opportunistic necrophagous scavengers (Collins et al. 2002), which could re-move the pressure of direct predation. The most studied and specious region of the Southern Ocean, South Georgia, have a rich endemic fauna despite the presences of a number of large durophagous predators of crab species with high abundances (Griffiths et al. 2009, Hogg et al. 2011, Barnes et al. 2009). This suggest that these crabs do not affect the number of species or habitat type in any harmful way, they instead form a part of a highly diverse commu-nity (Griffiths et al. 2013). A reason suggested why some of the crabs do not affect these communities, is because of their omnivorous behaviour, which is thought to distribute the predation pressure more and prevent elimination of any particular species (Britayev et al. 2010). Their omnivorous behaviour, or their spread of prey species, are reflected in their large numbers of prey species in the Weddell Sea compared to the native species, which is one of the overall things that determine their positions in the food web structure. This could thereby explain the low values the invasive crab species got in the PageRank analysis, and thus their lack of importance regarding the food web robustness.

There are less studies on how invasive durophagous sharks impact these types of ecosystems. In general sharks are important species in many of the ecolog-ical communities they are included in. They shape those marine ecosystems the live in, due to their roles as predators and scavengers they are essential to the health of these systems (Helfman & Burgess 2014). They are top-level carnivores, keeping abundances of other species in check, preventing overuse of resources and starvation, as well as remove sick and weak

indi-viduals. These predators also feed on remains of dead animals, which help the ecosystems with recycling of energy and nutrient (Helfman & Burgess 2014). This knowledge about sharks assume that they will make an impact on the food webs structure and the species in the Weddell Sea if they manage to invade. Affecting the abundances of their prey species and probably creating imbalances in this already complex ecosystem, which can propagate through the food chains. However, the results show that the invasive shark species do not impact the Weddell Sea food web in any considerable way. They end up with the lowest values in the PageRank centrality analysis, and are therefore not considered to be important in this food web regarding to it’s robustness. Sharks eat just about anything that is available, they are opportunists (Helf-man & Burgess 2014). Their opportunism on what to prey upon could work out in a similar way as the durophagous crabs around South Georgia, spread-ing the predation pressure and not affect abundances or habitat in any harmful way. Predation pressure and abundances were, however, factors which were not included in this study, and did not directly impact the PageRank centrality and were not the direct reason for the invasive species low PageRank val-ues. But there is a insinuation that these elements have an indirect effect on the outcome of the centrality analyses. The invasive species opportunism to-wards what to prey upon can be reflected by the large number of species they consume in the Weddell Sea food web. Together with their small or none ex-isting number of species who predate on them, places them in their current positions in the food web structure, and is probably a vital reason why these invasive species got such low values in the PageRank analysis.

PageRank centrality constantly serve as a high performer across real and hy-pothetical food webs among other centrality measures (McDonald-Madden et al. 2016). Other centrality metrics are often not adapted to food webs and how they work, which results in poor performances to evaluate species impor-tance compared to the PageRank centrality (Allesina & Pascual 2009). The similar result from the betweenness centrality, which is a similar model to the PageRank centrality, and the result from the sensitivity analyses, strengthen the result that these invasive species do not become important in this food web. What the reason could be for their lack of importance in the food web structure can only be speculated in. Most of the invasive species are highly connected, with numbers of interactions above the mean and median number of interactions per species in the Weddell Sea food web. And all of them

are predators, where almost all have a broad spectrum of species that they predate upon and a small or none existing number of predators. Their many prey species are probably the reason why they became so closely connected to many of the species in the food web, and the reason why they ended up in the top of the ranking order in the closeness centrality analysis. Despite their closeness, the invasive species are at the ends of their food chains within the food web, which suggest a lower flow of energy through them, and thereof probably their low scores in the betweenness and PageRank centrality analy-ses. Their opportunism regarding of what to predate upon, and thereby their connectedness and thus their position in the network, are probably the main reason why the invasive species got such low PageRank values and do not be-come important in this food web. The result might have turned out differently if the elements of predation pressure and abundances had been included in the study, together with the local impact of the invasive species.

It is difficult to know if large and propagating effects of invasion are general or not, due to the probability of bias studies. It is more presumable to observe spectacular effects since their effect can be seen directly (David et al. 2017), instead of research every food web on earth if they are invaded by new species and how they affect that web. Invasion of species do not always have to be negative and include extinction of species to a food web (David et al. 2017), it can actually increase species richness (Thomsen et al. 2014). Which in some way seems to be the case in the marine ecosystem surrounding South Georgia, where the presence of a number of invasive crab species do not affect the highly diverse community negatively, instead it thrives (Belchier et al. 2012). It has been shown that invaders can provide species at higher trophic levels than them self with habitat and food sources, which has resulted in positive effects on the diversity at higher trophic levels (Thomsen et al. 2014). Invasive species can also increase complexity and food chain length (Woodward & Hildrew 2001), as well as connectance and proportion of intermediate species (Salvaterra et al. 2013). Changes in food web structure because of invasion of species seem to rather depend on changes in abundances of trophic groups than extinction of species (Gallardo et al. 2016). This could be the very reason why the invasive species did not seem to make any significant impacts on the food web structure in this study, since abundances were not included in the data.

When impacts of invasive species lead to extinctions, it do not always act as the cause by its own. Their impact can be a part of simultaneous events, as environmental changes that co-occur, which drive species to extinction (Gurevitch & Padilla 2004). One example of this, is yet again the adult Nile Perch. Several of their prey species had already declined because of pollu-tion and over-harvesting before the rise of the Nile Perch (Downing et al. 2013, Goudswaard et al. 2008, Witte et al. 1992). Co-occurring environmen-tal changes will probably also occur in the Weddell Sea together with im-pacts from these invasive species, thought to invade, which could make the extinction risk higher. That part of the Southern Ocean experiencing one of Earth’s fastest rates of regional climate change. It’s surface temperature have increased more than 1 K in the last 60 years and the temperature of the deep water has risen as well, also the season of sea ice have become shorter (Clarke et al. 2007). These changes will have an impact on the complex food web of the Weddell Sea in addition to that of the invasive species. It is also possi-ble that the changes such as increased temperature will change how species do connect to each other within the food web, but those factors were not ac-counted for in this data set.

6

Conclusions

This study showed that the invasive species thought to invade the Weddell Sea food web due to rising water temperatures, do not alter the current structure of the food web in any considerable way. Nor do they them selves become important in the food web. The lack of changes in the food web structure due to the presents of these invasive species are probably a result of not in-cluding factors such as abundances and network dynamics in the analyses. Abundance seem to be one of the driving forces when it comes to changes in food web structure caused by invasion of species. To solely study food web structure through who eats whom does therefore not seem to be enough to de-termine the whole picture regarding the impact these invasive species has on the structure of the Weddell Sea food web. In areas nearby the Weddell Sea durophagous predators have already invaded and the native community have remained highly diverse. This is thought to be due to the invasive species

broad spectrum of prey. Their omnivorous behaviour and opportunism to-wards their food source is thought to distribute the predation pressure more and prevent elimination of any particular species. This suggest that the inva-sive species thought to invade the Weddell Sea will act similar in this com-munity. It is, however difficult to predict how the impact of these invasive species will co-occur with simultaneous events like environmental changes. Climate change which is likely to increase the water temperature and thereby also help the invasive species to invade, will probably also play an important part in how everything plays out in this ecosystem. Further climate change will most probably have a large impact on this fragile endemic fauna, which in combination with the impact of the invasive species could really alter the food web of the Weddell Sea.

7

Acknowledgements

First, I would like to thank my supervisors Anna Eklöf, György Barabás and Alyssa Cirtwill for all the help, guidance and advices you offered me during this project. I would also like to thank my course and office mates, Isabelle Norström and Mikael Ohlsson, for all the support, valuable comments and suggestions. Also huge thanks to my peers, Ulrica Ronquist and Raviv Gal, as well as my examiner, Karl-Olof Bergman, for all your comments and sug-gestions which made my work better. Finally, I would like to thank Anton Karlsson for the technical support with the programming and all answers to mathematical questions.

References

Ahyong ST & Dawson EW (2006). 1303 Lithodidae from the Ross Sea, Antarctica, with descriptions of two new species (Crustacea: Decapoda: Anomura). ZOOTAXA 1303, 45–68

Allesina S & Pascual M (2009). Googling food webs: Can an eigenvector measure species’ importance for coextinctions? PLoS Computational Bi-ology http://dx.doi.org/10.1371/journal.pcbi.1000494

Aronson RB, Thatje S, Clarke A, Peck LS, Blake DB, Wilga CD & Seibel BA (2007). Climate Change and Invasibility of the Antarctic Benthos. Annual Review of Ecology, Evolution, and Systematics http://dx.doi.org/ 10.1146/annurev.ecolsys.38.091206.095525

Aronson RB, Thatje S, Mcclintock JB & Hughes KA (2011). Anthropogenic impacts on marine ecosystems in Antarctica. Annals of the New York Academy of Sciences 1223, 82–107. http://dx.doi.org/10.1111/j.1749-6632.2010.05926.x

Barnes DKA, Griffiths HJ & Kaiser S (2009). Geographic range shift responses to climate change by Antarctic benthos: Where we should look. Marine Ecology Progress Series 393, 13–26. http://dx.doi.org/ 10.3354/meps08246

Belchier M, Peatman T & Brown J (2012). The biology, ecology and de-velopment of fishery management advice for the anomuran crabs at South Georgia (CCAMLR subarea 48.3)

Blackburn TM, Pyšek P, Bacher S, Carlton JT, Duncan RP, Jarošík V, Wilson JR & Richardson DM (2011). A proposed unified framework for biological invasions. Trends in Ecology and Evolution 26, 333–339. http://dx.doi.org/ 10.1016/j.tree.2011.03.023

Borrvall C, Ebenman B & Jonsson T (2000). Biodiversity lessens the risk of cascading extinctions in model food webs. Ecology Letters 3, 131–136. http://dx.doi.org/10.1046/j.1461-0248.2000.00130.x

Britayev TA, Rzhavsky AV, Pavlova LV & Dvoretskij AG (2010). Stud-ies on impact of the alien Red King Crab (Paralithodes camtschaticus)

on the shallow water benthic communities of the Barents Sea. Jour-nal of Applied Ichthyology 26, 66–73. http://dx.doi.org/10.1111/j.1439-0426.2010.01494.x

Brose U, Ostling A, Harrison K & Martinez ND (2004). Unified spatial scaling of species and their trophic interactions. Nature 428, 167–171. http://dx.doi.org/10.1038/nature02297

Bruno JF, Fridley JD, Bromberg KD & Bertness MD (2005). Insights into Bi-otic Interactions from Studies of Species Invasions. Insights into Ecology, Evolution, and Biogeography pp. 13–40

Clarke A, Murphy EJ, Meredith MP, King JC, Peck LS, Barnes DK & Smith RC (2007). Climate change and the marine ecosystem of the western Antarctic Peninsula. Philosophical Transactions of the Royal Society B: Bi-ological Sciences 362, 149–166. http://dx.doi.org/10.1098/rstb.2006.1958

Collins MA, Yau C, Guilfoyle F, Bagley P, Everson I, Priede IG & Agnew D (2002). Assessment of stone crab (Lithodidae) density on the South Georgia slope using baited video cameras. ICES Journal of Marine Science 59, 370–379. http://dx.doi.org/10.1006/jmsc.2001.1167

David P, Thébault E, Anneville O, Duyck PF, Chapuis E & Loeuille N (2017). Impacts of Invasive Species on Food Webs: A Review of Em-pirical Data. Advances in Ecological Research 56, 1–60. http://dx.doi.org/ 10.1016/BS.AECR.2016.10.001

Downing AS, Galic N, Goudswaard KPC, van Nes EH, Scheffer M, Witte F & Mooij WM (2013). Was Lates Late? A Null Model for the Nile Perch Boom in Lake Victoria. PLoS ONE 8. http://dx.doi.org/ 10.1371/journal.pone.0076847

Dunne JA, Williams RJ & Martinez ND (2002). Network structure and biodi-versity loss in food webs: robustness increases with connectance. Ecology Letters 5, 558–567. http://dx.doi.org/10.1046/j.1461-0248.2002.00354.x

Ebenman B, Law R & Borrvall C (2004). Community viability analysis: The response of ecological communities to species loss. Ecology 85, 2591– 2600. http://dx.doi.org/10.1890/03-8018

Eklöf A & Ebenman B (2006). Species loss and secondary extinctions in simple and complex model communities. Journal of Animal Ecology 75, 239–246. http://dx.doi.org/10.1111/j.1365-2656.2006.01041.x

Estes JA & Palmisano JF (1974). Sea Otters: Their Role in Structuring Nearshore Communities. Science 185, 1058–1060

FIelds PA, Graham JB, Rosenblatt RH & Somero GN (1993). Effects of expected global climate change on marine faunas. Trends in Ecology and Evolution 8, 361–367. http://dx.doi.org/10.1016/0169-5347(93)90220-J

Gallardo B, Clavero M, Sánchez MI & Vilà M (2016). Global ecological impacts of invasive species in aquatic ecosystems. Global Change Biology 22, 151–163. http://dx.doi.org/10.1111/gcb.13004

Gardner MR & Ashby WR (1970). Connectance of Large Dynamic (Cyber-netic) Systems: Critical Values for Stability. Nature 228

Goudswaard K, Witte F & Katunzi EF (2008). The invasion of an introduced predator, Nile perch (Lates niloticus, L.) in Lake Victoria (East Africa): Chronology and causes. Environmental Biology of Fishes 81, 127–139. http://dx.doi.org/10.1007/s10641-006-9180-7

Griffiths HJ, Barnes DK & Linse K (2009). Towards a generalized biogeogra-phy of the Southern Ocean benthos. Journal of Biogeograbiogeogra-phy 36, 162–177. http://dx.doi.org/10.1111/j.1365-2699.2008.01979.x

Griffiths HJ, Whittle RJ, Roberts SJ, Belchier M & Linse K (2013). Antarctic Crabs: Invasion or Endurance? PLoS ONE 8. http://dx.doi.org/10.1371/

Gurevitch J, Fox GA, Wardle GM, Inderjit & Taub D (2011). Emergent insights from the synthesis of conceptual frameworks for biological in-vasions. Ecology Letters 14, 407–418. http://dx.doi.org/10.1111/j.1461-0248.2011.01594.x

Gurevitch J & Padilla DK (2004). Are invasive species a major cause of ex-tinctions? Trends in Ecology and Evolution 19, 470–474. http://dx.doi.org/ 10.1016/j.tree.2004.07.005

Helfman G & Burgess GH (2014). Sharks: The Animal Answer Guide. Johns Hopkins University Press, Baltimore

Helmuth B, Mieszkowska N, Moore P & Hawkins SJ (2006). Living on the Edge of Two Changing Worlds: Forecasting the Responses of Rocky Intertidal Ecosystems to Climate Change. Annual Review of Ecology, Evolution, and Systematics 37, 373–404. http://dx.doi.org/ 10.1146/annurev.ecolsys.37.091305.110149

Hogg OT, Barnes DK & Griffiths HJ (2011). Highly diverse, poorly stud-ied and uniquely threatened by climate change: An assessment of ma-rine biodiversity on South Georgia’s continental shelf. PLoS ONE 6. http://dx.doi.org/10.1371/journal.pone.0019795

Intergovernmental Panel on Climate Change (2014). Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change

Jacob U (2005). Trophic Dynamics of Antarctic Shelf Ecosystems-Food Webs and Energy Flow Budgets, Universität Bremen. Universität Bremen. Home| About ePIC| Impressum| . . .

Jeschke JM, Bacher S, Blackburn TM, Dick JT, Essl F, Evans T, Gaertner M, Hulme PE, Kühn I, Mrugała A, Pergl J, Pyšek P, Rabitsch W, Ricciardi A, Richardson DM, Sendek A, Vilà M, Winter M & Kumschick S (2014). Defining the impact of non-native species. Conservation Biology 28, 1188– 1194. http://dx.doi.org/10.1111/cobi.12299

Koh LP, Dunn RR, Sodhi NS, Colwell RK, Proctor HC & Smith VS (2004). Species Coextinctions and the Biodiversity Crisis. Supporting Online Ma-terial. Science 305, Supporting–Online Material

Levin SA (1992). The Problem of Pattern and Scale in Ecology. Ecology 73, 1943–1967. http://dx.doi.org/doi:10.2307/1941447

May RM (1986). The Search for Patterns in the Balance of Nature: Advances and Retreats. Ecology 67, 1115–1126

McDonald-Madden E, Sabbadin R, Game ET, Baxter PW, Chadès I & Poss-ingham HP (2016). Using food-web theory to conserve ecosystems. Nature Communications 7. http://dx.doi.org/10.1038/ncomms10245

Murtaugh PA & Derryberry DR (1998). Models of Connectance in Food Webs 54, 754–761

Newman MEJ (2003). The Structure and Function of Complex Networks. SIAM Review 45, 167–256. http://dx.doi.org/10.1137/S003614450342480

Newman MEJ (2010). Networks. An introduction

Paine RT (1966). Food Web Complexity and Species Diversity. The American Naturalist 100, 65–75

Pantel JH, Bohan DA, Calcagno V, David P, Duyck PF, Kamenova S, Loeuille N, Mollot G, Romanuk TN, Thébault E, Tixier P & Massol F (2017). 14 Questions for Invasion in Ecological Networks. Advances in Ecological Research 56, 293–340. http://dx.doi.org/10.1016/bs.aecr.2016.10.008

Pascual M & Dunne JA (2006). Ecological networks : linking structure to dynamics in food webs. Oxford University Press

R Core Team (2016). R: A language and environment for statistical comput-ing. Vienna, Austria

Salvaterra T, Green DS, Crowe TP & O’Gorman EJ (2013). Impacts of the invasive alga Sargassum muticum on ecosystem functioning and food web structure. Biological Invasions 15, 2563–2576. http://dx.doi.org/ 10.1007/s10530-013-0473-4

Schoener TW (1989). Food Webs From the Small to the Large. Ecology 70, 1559–1589

Smetacek V & Nicol S (2005). Polar ocean ecosystems in a changing world. Nature 437, 362–368. http://dx.doi.org/10.1038/nature04161

Smith CR, Grange LJ, Honig DL, Naudts L, Huber B, Guidi L & Domack E (2012). A large population of king crabs in Palmer Deep on the west Antarctic Peninsula shelf and potential invasive impacts. Proceedings of the

Royal Society B: Biological Sciences 279, 1017–1026. http://dx.doi.org/ 10.1098/rspb.2011.1496

Sole RV & Montoya M (2001). Complexity and fragility in ecological net-works. Proceedings of the Royal Society B: Biological Sciences 268, 2039– 2045. http://dx.doi.org/10.1098/rspb.2001.1767

Stachowicz JJ, Terwin JR, Whitlatch RB & Osman RW (2002). Linking climate change and biological invasions: Ocean warming facilitates non-indigenous species invasions. Proceedings of the National Academy of Sciences of the United States of America 99, 15497–500. http://dx.doi.org/ 10.1073/pnas.242437499

Thatje S (2005). The future fate of the Antarctic marine biota?Thatje, S. (2005). The future fate of the Antarctic marine biota? Trends in Ecology and Evolution, 20(8), 418–419. https://doi.org/10.1016/j.tree.2005.05.006. Trends in Ecology and Evolution 20, 418–419. http://dx.doi.org/ 10.1016/j.tree.2005.05.006

Thatje S, Anger K, Calcagno JA, Lovrich GA, Pörtner HO & Arntz WE (2005). Challenging the cold: Crabs reconquer the antarctic. Ecology 86, 619–625. http://dx.doi.org/10.1890/04-0620

Thatje S & Lörz AN (2005). First record of lithodid crabs from Antarctic wa-ters off the Balleny Islands. Polar Biology 28, 334–337. http://dx.doi.org/ 10.1007/s00300-004-0686-1

Thomsen MS, Byers JE, Schiel DR, Bruno JF, Olden JD, Wernberg T & Sil-liman BR (2014). Impacts of marine invaders on biodiversity depend on trophic position and functional similarity. Marine Ecology Progress Series 495, 39–47. http://dx.doi.org/10.3354/meps10566

Walther GR, Post E, Convey P, Menzel A, Parmesan C, Beebee TJC, Fro-mentin JM, Hoegh-Guldberg O & Bairlein F (2002). Ecological re-sponses to recent climate change. Nature 416, 389–395. http://dx.doi.org/ 10.1038/416389a

Witte F, Goldschmidt T, Wanink J, van Oijen M, Goudswaard K, Witte-Maas E & Bouton N (1992). The destruction of an endemic species flock:

quantitative data on the decline of the haplochromine cichlids of Lake Victoria. Environmental Biology of Fishes 34, 1–28. http://dx.doi.org/ 10.1007/BF00004782

Woodward G & Hildrew AG (2001). Invasion of a stream food web by a new top predator. Journal of Animal Ecology 70, 273–288. http://dx.doi.org/ 10.1046/j.1365-2656.2001.00497.x

8

Appendix

Figure 8: Sensitivity analysis for the Weddell Sea food web including the invasive species, where 5 % of the interactions were randomly removed. The black dots represent the original Pagerank value, when no interactions were removed. The figure shows the 19 species with the originally highest ranked Pagerank values.

Figure 9: Sensitivity analysis for the Weddell Sea food web including the invasive species, where 15 % of the interactions were randomly removed. The black dots represent the original Pagerank value, when no interactions were removed. The figure shows the 19 species with the originally highest ranked Pagerank values.