Spatial complexity and fit between ecology and management: Making sense of patterns in fragmented landscapes

(1)

Spatial complexity and ﬁt between ecology and management Making sense of patterns in fragmented landscapes

Arvid Bergsten

Doctoral thesis in Natural Resource Management Stockholm Resilience Centre

Stockholm 2013

(2)

Doctoral dissertation 2014 Stockholm Resilience Centre Stockholm University SE-‐106 91 Stockholm, Sweden

ISBN 978-‐91-‐7447-‐834-‐1, pages 1-‐30 Cover illustration: Concetta Flore

Printed by Universitetsservice US-‐AB in Stockholm 2013

Paper 1, 2 and 3 are reprinted with permission from the publisher

(3)

Abstract

Avoiding the negative eﬀects of habitat fragmentation on biodiversity is especially challenging when also the management institutions are spatially and administratively distributed. This doctoral thesis introduces ﬁve case studies that investigate ecological, social and social-‐ecological relations in frag-‐

mented landscapes. I present new approaches in which research and governance can detect and man-‐

age mismatches between landscape ecology and planning. The case studies include urban and forested landscapes where an intense land-‐use is limiting the connectivity, i.e., the potential for many species to disperse between the remaining patches of habitat. Graph-‐theoretic (network) models are applied to map connectivity patterns and to estimate the outcome for dispersing species at the patch level and for the whole study system. In particular, the network models are applied to evaluate the spatial complex-‐

ity and the potential mismatches between ecological connectivity and geographically distributed man-‐

agement institutions like protected areas and municipalities. Interviews with municipal ecologists complement the spatial analysis; revealing some problems and ways forward regarding the communi-‐

cation and integration of ecological knowledge within local spatial-‐planning agencies. The results also show that network models are useful to identify and communicate critical ecological and social-‐

ecological patterns that call for management attention. I suggest some developments of network mod-‐

els as to include interactions between species and across governance levels. Finally, I conclude that more eﬀort is needed for network models to materialize into ecological learning and transformation in management processes.

Keywords: Connectivity; Conservation; Dispersal; Ecological knowledge; Ecology; Forest; Fragmenta-‐

tion; Graph theory; Institutional ﬁt; Landscape; Management; Metapopulation; Municipal ecologist;

Network; Planning; Protected area; Scale mismatch; Social-‐Ecological; Urban; Wetland

(4)

Populärvetenskaplig sammanfattning

Växelverkan mellan människan och andra arter skapar komplexa mönster i landskapet. Arters utbred-‐

ning beror på var de ﬁnner livsmiljöer (habitat) och spridningsmöjligheter. Spridning mellan habitat-‐

ﬂäckar skapar mönster av "konnektivitet", vilket avgör var vi påträﬀar många arter och som är nödvän-‐

digt för arters långsiktiga överlevnad i fragmenterade landskap. Människan omformar konnektiviteten när vi bygger städer eller skördar mat och material. Dessa aktiviteter är utspridda, genom människors direkta användning av landskapet, och indirekt genom samarbete mellan t ex kommuner, bönder eller skogsägare. Förlusten och fragmenteringen av sammanhängande livsmiljöer till avskilda habitatﬂäckar är det största hotet mot världens biologiska mångfald och tillgång på ekosystemtjänster. Vi måste där-‐

för se till att de livsmiljöer som återstår – eller som återskapas – utnyttjar den begränsade ytan på ett sätt som är rumsligt eﬀektivt. Detta innebär att naturresurserna hålls tillgängliga både för människan och för arter som behöver spridas för att överleva.

I doktorsavhandlingen "Spatial complexity and ﬁt between ecology and management: Making sense of patterns in fragmented landscapes" studeras olika samband mellan landskapsplanering och arters kon-‐

nektivitet i fragmenterade landskap. Denna forskning bygger på att hotade arter kan skyddas om vi tar fram verktyg för att kartlägga platser och strukturer i landskapet som är kritiska för arternas spridning, och sedan tillämpar sådan kunskap i landskapsplaneringen. Författaren presenterar fem uppsatser som studerar såväl enskilda arter som artgrupper, och som undersöker landskapsförvaltning i form av om-‐

rådesskydd och kommuners fysiska planering. Avhandlingen tar fram ny kunskap som är användbar i både forskning och planering, samt utforskar hinder och metoder för ett ökat utbyte mellan de två sfä-‐

rerna.

I ett norrländskt skogslandskap tillämpas nätverksmodeller av konnektivitet för att förklara förekoms-‐

ten av sorkar och för att utvärdera hur produktionsskog och olika områdesskydd kompletterar varand-‐

ra. I Stockholmsregionen tillämpas en nätverksmodell av våtmarker och mellankommunala samarbe-‐

ten. Baserat på en arts spridningsavstånd och avståndet mellan habitatﬂäckar, kan nätverksanalys påvi-‐

sa kritiska områden samt jämföra olika landskap eller scenarier. Nätverksmodellerna tillämpas i analy-‐

sen på nya sätt för att undersöka hur konnektiviteten klaﬀar med planeringsområden i det "administra-‐

tiva landskapet". Problem kan uppstå till exempel om funktionen hos en våtmark i kommun A är bero-‐

ende av konnektivitet till våtmarker i kommun B, fast kommunerna saknar dialog om våtmarker och om sin respektive planering. Resultaten visar också att vissa arter kan nå fram till värdefulla skogar i naturreservat genom att sprida sig via nyckelbiotoper och äldre produktionsskogar.

Nätverksanalyserna kompletteras av intervjuer med kommunekologer om hur ekologisk kunskap kan produceras och kommuniceras bättre inom fysisk planering. Resultaten visar att kommunerna saknar systematiska verktyg för att förstå och hantera konnektivitet i landskapet, särskilt över kommungrän-‐

ser där mer samarbete krävs. Nätverksmodellen sågs som ett kraftfullt verktyg av tjänstemän som hade testat ett datorprogram skräddarsytt för planering. Vidare visar intervjuerna vilka områden av ekolo-‐

gisk kunskap som kommunekologerna framhäver i planeringsdiskussioner. Lättast har de att förmedla naturens betydelse för människan och minst förståelse möter de för sin kunskap om arter och ekosy-‐

stemens dynamik och komplexitet. Intervjuresultaten betonar att planeringens lärandeprocesser och anpassning till det lokala ekosystemet förbättras om kommunekologer verkar inom planeringsgrup-‐

perna, i motsats till när "färdigproducerad" kunskap tas in utifrån, inte minst vad gäller nätverksanaly-‐

ser av arters spridningsmöjligheter. Slutligen föreslås ny forskning om hur nätverksmodellen kan in-‐

kludera interaktioner mellan arter och mellan aktörer på lokal och regional nivå.

(5)

List of papers

1. Magnusson M, Bergsten A, Ecke F, Bodin Ö, Bodin L & Hörnfeldt B (2013). Predicting grey-‐

sided vole occurrence in northern Sweden at multiple spatial scales. Ecology and Evolution 3:4365–4376.

2. Bergsten A, Bodin Ö & Ecke F (2013). Protected areas in a landscape dominated by logging – A connectivity analysis that integrates varying protection levels with competition–colonization tradeoﬀs. Biological Conservation 160:279–288.

3. Bergsten A & Zetterberg A (2013). To model the landscape as a network: A practitioner’s per-‐

spective. Landscape and Urban Planning 119:35–43.

4. Bergsten A. Communicating ecology in local planning – the role of embedded ecologists. Manu-‐

script.

5. Bergsten A, Galafassi D & Bodin Ö. The problem of ﬁt in social-‐ecological systems: Detecting spatial mismatches between ecological connectivity and land management in an urban region.

In review in Ecology and Society.

My contributions to the papers

For all papers I developed the research design and performed the analysis in collaboration with the co-‐

authors. I led the writing of all papers except paper 1, to which I contributed mainly with the habitat mapping, ﬁeldwork, connectivity analysis and manuscript preparation.

(6)

1. Introduction

Systems approaches such as ecosystem-‐based man-‐

agement state that natural resource management should recognize interactions in ecosystems, rather than focusing on single components in isolation (e.g., Christensen et al. 1996). A species’ role in an ecosys-‐

tem can be seen in several dimensions, for example, through its interaction with other species in a local community (e.g., Montoya et al. 2006; Dyer et al.

2010); through its interactions with the abiotic envi-‐

ronment (e.g., Chapin & Shaver 1996; Jones et al.

1994; Lawrence et al. 2012); or through its importance for human societies (e.g., Cardinale et al. 2012; Con-‐

stanza et al. 1997; Naeem et al. 2009). Another di-‐

mension is the species distribution, and how this de-‐

pends on the species’ interaction with the distribu-‐

tion of resource (habitat) patches. The spatial con-‐

ﬁguration of habitat aﬀects the functioning of indi-‐

vidual patches and the local species populations liv-‐

ing there, as well as it aﬀects the system-‐level proper-‐

ties on a metapopulation, landscape or seascape level (Hanski 1998). A greater distance between two patches generally means a weaker connection in terms of organism movement, ﬂows of energy and nutrients, exchange of genetic material, and other processes.

Ecological processes across space create a connec-‐

tivity pattern whose complexity increases exponen-‐

tially with the spatial scale of observation (Levin 1992). Connectivity is thus the outcome of the inter-‐

action between an organism’s behavior and the spa-‐

tial distribution of relevant landscape features (Taylor et al. 1993). A patch-‐connectivity perspective can support the evaluation of ecosystem complexity and of structure–function relationships (Fortin et al. 2003;

Pickett et al. 2005). Important interactions exist on speciﬁc spatial (or temporal) scales as well as across scales (e.g. Holling 1992; Olﬀ & Ritchie 2002). Spatial processes have, just like local multitrophic interac-‐

tions, the potential to generate nonlinear eﬀects on biodiversity and on ecosystem functioning . Cross-‐

scale interactions involve elements or functions on more than one level of ecological organization, which may produce unexpected eﬀects at scales larger or smaller than the observation scale. For example, the

loss of a few critical habitat patches may cause a spe-‐

cies to vanish from an entire landscape, and a slight general population decline may produce local extinc-‐

tions.

1.1. The spatial relation of ecology and governance patterns

In order to have the capacity to avoid harmful spatial mismatches, ecosystem-‐based management must A) operate on a spatial scale large enough to deal with those ecological processes and functions that are con-‐

sidered important, and B) be capable to learn about such processes and the related cross-‐scale interac-‐

tions in a way that is useful to the governance system and to society. Hence, applying a large-‐scale perspec-‐

tive is often necessary, but with large scales comes the problem that spatial compositions become more diﬃcult to understand and manage, as the potential complexity skyrockets.

In addition, the number of potential spatial interac-‐

tions multiplies when we take into account the ar-‐

rangement of spatially distributed actors. Actors are often interconnected by various socioeconomic proc-‐

esses (e.g., Pickett et al. 2005); and linked to natural resources, as emphasized by the social-‐ecological-‐

systems perspective (SES; Folke 2006). The manage-‐

ment of most ecosystems operates within the spatial subdivisions delimited primarily by the jurisdictions from local up to multinational governance levels, and secondarily by the individual actors' planning and implementation of management actions within their respective areas of operation. Mismatches between ecological and administrative patterns lead to prob-‐

lems for the social system responsible for manage-‐

ment and/or for the ecological systems being man-‐

aged (Berkes 2006; Borowski et al. 2008; Brondizio et al. 2009; Brown 2003; Cumming et al. 2006; Folke et al. 1997, 2007; Guerrero et al. 2013; Young 2002). It may cause local loss of system elements and disrupt or inhibit system functions, like self-‐maintenance, nutrient cycles, dispersal processes or provision of ecosystem services (Cumming et al. 2006; Levin 1999).

(8)

1.2. Understanding mismatches through spatially explicit models

This thesis recognizes that long-‐term solutions to spatial mismatches will depend on the capacity of managing actors and agencies to identify spatially explicit ecological dependencies (patterns and proc-‐

esses), to understand their causes and consequences, and to accordingly adjust their policy and manage-‐

ment practices at appropriate scales. During my PhD I have received response from practitioners at local, regional and national governance agencies (mainly in Sweden) and from researchers and forest companies;

indicating a large and relatively unrealized learning potential in heuristic modeling tools that help institu-‐

tions to learn about spatial ecological patterns, and to manage them so harmful social-‐ecological mis-‐

matches can be avoided.

To be eﬀective for learning and management, I argue, such model should fulﬁll two criteria. It should 1) rep-‐

resent ecological connectivity in a valid way and 2) promote cognition and/or communication for the practitioners or researchers that apply the model or use the results. System models typically exhibit a general tradeoﬀ between the two criteria. While the inclusion of more variables, interactions and dynam-‐

ics can achieve a more realistic model of system be-‐

havior; it also involves additional sources of errors and biases and a more costly estimation of parame-‐

ters and state variables (Ascough et al. 2008; Perz et al. 2013). Furthermore, a more complex model usually demands a higher competence for using the model, and it tends to be less versatile in terms of robustness and applicability to diﬀerent case contexts. In con-‐

trast, reducing the model complexity may make the model more flexible to varying cases, and the outputs more intelligible to diverse users. However, a simpler model may fail to adequately predict and represent the system due to the simplifications of key processes that affect real-‐world complexity (Nihoul 1994). Find-‐

ing this model balance is even more diﬃcult for ap-‐

plication in multilayered governance system. For ex-‐

ample, while models used in conservation planning at the regional scale may eﬀectively deal with comple-‐

mentarity and connectivity of protected areas, they often fail to inform local management actions in a

useful way (Mills et al. 2010). Hence, reﬂecting over the spatial scales of policy and practice is essential for the development of socially functional ecological models.

This thesis speciﬁcally explores the potential of con-‐

nectivity graph (abbreviated “CG” in this text; the graph-‐theoretic model of spatial connectivity is ex-‐

plained in the methods chapter). It is argued that this type of model balances the tradeoﬀ between criteria 1 and 2 above in a way that captures spatial ecological complexity relevant for evaluating management sce-‐

narios (Bodin & Norberg 2007; Bunn et al. 2000; Cal-‐

abrese & Fagan 2004; Keith et al 1997; Laita et al. 2011;

Minor & Urban 2008; Pascual-‐Hortal & Saura 2006;

Rayﬁeld et al. 2011.) The CG provides a quantitative modeling framework to evaluate ecological function at multiple levels of ecological organization, from nodes (e.g., patches) to graph components (groups of connected patches, to the whole graph (all patches in the study system). It is a robust model given the rela-‐

tively low demands on data (Urban et al. 2009). This thesis will later outline how I used the CG to evaluate ecological and social-‐ecological spatial patterns, in three large study areas comprising up to 20,000 habi-‐

tat patches.

1.3. Connecting landscape research and management

My PhD research builds on the belief that landscape research and management mutually beneﬁt from in-‐

teracting with each other. Research helps landscape management and planning to sustain ecological func-‐

tioning by developing applied and conceptual knowl-‐

edge and assessment tools. Management practice helps academic research by signaling problems, by evaluating the usability of solutions proposed by re-‐

searchers, and by helping research to pose questions important beyond academia. However, the diﬀerent agendas that drive the two spheres do not always stimulate interaction. Practitioners generally concen-‐

trate on meeting the demands of end users such as urban dwellers (in the case of urban planning) or for-‐

est owners (in the case of forestry), while simultane-‐

ously trying to take in advice regarding landscape

(9)

ecology and sustainability in general from scientists and policy-‐makers. Landscape researchers, on the other hand, often concentrate on the ecological func-‐

tion of a species or a local community, with pristine environments in focus or as benchmark, disregarding the human domination of most landscapes (Ellis &

Ramankutty 2008). The papers in this thesis seek to contribute to the research as well as to the manage-‐

ment of landscapes, by developing and applying ap-‐

proaches that connects the two spheres.

1.4. Aim and research questions

The aim of this thesis is to improve the understand-‐

ing of complexity of social-‐ecological landscapes, in a way that advances the research as well as the man-‐

agement of landscapes, and that facilitates the inter-‐

action between the two spheres. In paper 1, 2 and 5 I use connectivity graphs (CGs) to capture complexity with respect to ecological interactions and to admin-‐

istrative arrangements, with the intention to produce results useful and intelligible to management institu-‐

tions, but also to point out some needs for further development of spatially explicit ecosystem models.

Paper 3 and 4 are based on interviews with practitio-‐

ners in which I investigate the factors that make eco-‐

logical knowledge useful in municipal spatial-‐

planning practice; in a way that I hope contributes to the salience and adaption of landscape-‐ecological knowledge in decision and policy making. To sum up, the ﬁve papers connect the three thesis questions on the right hand side.

Scope of papers

Table 1 shows how the papers focus diﬀerent thesis questions. Paper 1 tackles Q1 by studying connectivity in combination with other ecological factors that hy-‐

pothetically predict occurrence of the grey-‐sided vole (Myodes rufocanus) in Swedish boreal forests. Paper 2 links the concept of habitat connectivity to the spe-‐

cies' tradeoﬀ in colonization–competition ability, through diﬀerentiating the ways whereby mature for-‐

est patches contribute to the total connectivity of a habitat network. Paper 2 also addresses Q3 by the

way our analysis diﬀerentiates between forest patches prioritized for wood production, versus for biodiver-‐

sity through multiple area-‐protection institutions.

Paper 3 approaches Q2 by studying the consideration of ecological connectivity in the spatial-‐planning process of municipalities in the Stockholm region, and by examining the pros and cons that practitio-‐

ners see with using a CG software tool to assess and communicate connectivity in planning discussions.

Paper 4 investigates these municipal ecologists' tech-‐

niques and diﬃculties when trying to integrate di-‐

verse ecological knowledge in the planning process.

Paper 5 researches Q3 by localizing mismatches ay municipal boundaries between wetland connectivity and collaboration regarding wetland management.

Table 1. Main focus of the ﬁve papers.

Thesis Ques)ons Thesis Ques)ons Thesis Ques)ons

Q1 Q2 Q3

Paper 1 Paper 2 Paper 3 Paper 4 Paper 5

Q1. How can CGs further our understanding of diverse ecological processes, in a way that is meaningful for conservation and landscape planning?

Q2. How is ecological knowledge in general and in particular about connectivity communi-‐

cated by municipal practitioners and re-‐

ceived in the local planning process?

Q3. How can CG models of the match between ecological connectivity and spatially dis-‐

tributed governance institutions enhance the knowledge about social-‐ecological ﬁt and improve landscape planning?

(10)

2. Background

This chapter provides some theoretical and empirical background to the papers presented in chapter 4.

2.1. Connectivity and landscape change

Several of the papers study ecological connectivity.

This section introduces some diﬀerent perspectives on connectivity and related research ﬁelds.

Land-‐use change drives habitat fragmentation As the world’s population has increased we have seen an intensiﬁed use of land and water areas. Urbaniza-‐

tion and the increased production of food and mate-‐

rials drives landscape fragmentation, which alters abiotic processes (Saunders et al. 1991) and increases the loss and isolation of many habitat types (Bogaert et al. 2005). As a response, 92 countries so far have signed the Nagoya protocol aiming to decrease the loss of ecosystem functions and services by protecting at least 17% of their respective land area in “well con-‐

nected systems” (in combination with other meas-‐

ures; Aichi Biodiversity Target #11, CBD Secretariat 2010). Many of these ecosystem services are sustained by species whose survival, reproduction and mobility are impinged by habitat fragmentation, i.e., the loss of connectivity between resource patches (Debinski &

Holt 2000; Ewers & Didham 2006; Fahrig 2003; Frank-‐

lin et al. 2002; Foley et al. 2005; Lindenmayer &

Fischer 2006; Pimm & Raven 2000). The risk of re-‐

gional species extinction accelerates when the habitat area fall below 10-‐30% of the historical coverage (de-‐

pending on the species' area requirements; Andrén 1994, 1997; cf. Betts 2006; Fahrig 2002; Radford et al.

2005).

Connectivity and landscape planning

What spatial conﬁgurations of habitat patches are then desirable given a limited total area of habitat?

Connectivity is commonly deﬁned as the degree to which the landscape facilitates or impedes the movement of species among resource patches and is

thus an outcome of an organism’s behavior in inter-‐

action with the distribution of landscape features (Merriam 1984; Taylor et al. 1993; Wiens 1989). Con-‐

nectivity can be managed and protected by prioritiz-‐

ing the conservation or restoration of habitat at sites that are critical to connectivity at the system scale, for example, when allocating a limited area for con-‐

servation or when compensating for connectivity losses due to development or climate change (Carroll et al. 2004; Crooks & Sanjayan 2006; Hanski 2011;

Heller & Zavaleta 2009; Huxel & Hastings 1999; Gur-‐

rutxaga et al. 2011). Hence, while combating habitat loss is the principal challenge for biodiversity conser-‐

vation, a simultaneous and well-‐informed manage-‐

ment of connectivity will increase the cost-‐beneﬁt ratio of conservation actions (Pressey & Bottrill 2009;

cf. Hodgson et al. 2009; Yaacobi et al. 2007).

Metapopulation theory

In landscape research, the understanding of connec-‐

tivity has developed around metapopulation theory, which was initiated by Levins (1969) and advanced by Hanski (1991, 1998). A metapopulation consists of local populations located in separate but connected habitat patches. Most local populations are so small that they will likely disappear due to stochastic events of local extinction, if it were not for the spatial con-‐

text that enables local extinctions to be compensated for through colonization from local populations in other patches. Metapopulation theory predicts that a species will persist in a landscape or seascape if the recolonization rate compensates for the rate of ran-‐

dom local extinctions (Hanski 1998). The rate of colonization between two local populations generally depends on 1) the distance between and 2) the areas of the two habitat patches (Fig 1; Hanski & Ovaska-‐

inen 2000). The patch area (possibly weighted by habitat quality) determines the maximum local popu-‐

lation size, which determines the rate of emigrating individuals. A larger patch is also more likely to re-‐

ceive colonizers, which inﬂuences the immigration rate. A larger distance decreases the colonization rate between two patches. However, a dispersing individ-‐

ual may experience the distance as shorter or longer depending on the land-‐cover between the patches,

(11)

which is referred to as the ”matrix quality” (Kupfer et al. 2006; Vandermeer & Carvajal 2001).

Fig 1. Migration to and from a subpopulation in the left patch. Letter d represents the interpatch distance (Euclid-‐

ean or land-‐cover weighted), and a represents the patch attribute (area and/or habitat quality).

Dynamic landscapes

In a social-‐ecological system, local extinctions are caused not only by random population dynamics but also by human drivers like urbanization, agriculture or forestry. Anthropogenic disturbance and fragmen-‐

tation increase the role of dispersal for species persis-‐

tence (Bengtsson 2010). On the other hand, human drivers also make new habitat patches available, for example through habitat restoration projects, or as forest stands regrow after logging. These activities are spatially distributed, aﬀecting speciﬁc habitat patches and leaving the remaining ones intact. Also, most activities are planned in time and space; by individu-‐

als, associations or authorities. The connectivity pat-‐

tern is therefore changing dynamically as a result of both planned an unplanned “disturbances”. The spa-‐

tial resilience of a metapopulation in a dynamic land-‐

scape or seascape can be deﬁned as its ability to re-‐

cover and persist over time in the face of spatially distributed disturbances (cf. Bengtsson et al. 2003;

Cumming 2011; Nyström & Folke 2001). In this con-‐

text, landscape research can be made useful to land-‐

scape management and conservation, by linking pat-‐

terns of connectivity and disturbances to the viability of species in dynamic landscapes (Cabeza & Moilanen 2003; Teeﬀelen et al. 2012).

d Emigration rate = f(d, a₁,a₂)

Immigration rate = f(d, a₁,a₂)

a

₂

a

₁

The graph model of connectivity

A spatial-‐planning process will increase or decrease the spatial resilience of fragmented metapopulations, depending on how knowledge about connectivity is produced, heeded and applied. Science has developed connectivity models useful to compare spatial scenar-‐

ios or to evaluate the consequences of speciﬁc land-‐

use changes. A recent model is the graph-‐theoretical model of connectivity, which is implemented in many quantitative measures of connectivity, such as the IIC measure described in the methods chapter. As speci-‐

ﬁed by the metapopulation model, the IIC estimates connectivity between two habitat patches as a prod-‐

uct of on 1) the distance between and 2) the areas of the two patches.

Landscape ecology and in particular metapopulation models often represent habitat areas as patches in a surrounding non-‐habitat matrix (Forman 1995). In a connectivity graph, nodes represent patches, con-‐

nected by links that represent the potential for an organism to move between two patches, for example during dispersal (Fig. 2; Dale & Fortin 2010; Urban et al. 2009). Nodes are connected by links only when the distance between two patches does not exceed the maximum dispersal distance for the model organism.

Most applications of network analysis have used geo-‐

graphic distance (Galpern et al. 2011), whereas others have applied traversal-‐cost thresholds based on the landscape permeability experienced by a species when dispersing through diﬀerent types of land cover in the matrix (e.g., Bunn et al. 2000). Hence, links are not interpreted as structural features of the landscape such as corridors, but as functional connections be-‐

tween patches as a dispersing organism might experi-‐

ence them. Additional information about a species’

biology and empirical observations can be used to validate or calibrate existing network models (e.g., Andersson & Bodin 2009; Fall et al. 2007; Minor &

Urban 2008). Critical connectivity thresholds can be identiﬁed in models by systematically removing nodes or links (e.g., Bascompte et al. 2006; Bodin et al. 2006; Brooks 2006; Saura & Rubio 2010).

(12)

Fig. 2. All the nodes and links in the image make up the graph. An organism inhabiting a node within a network component can disperse to other nodes in the same compo-‐

nent, although the dispersal probability decreases if several intermediate nodes need to be traversed (cf. Pascual-‐Hortal

& Saura 2006). Dark gray represents high betweenness cen-‐

trality, i.e., that a patch is identiﬁed as a stepping-‐stone for organisms dispersing between other patches.

2.2. Empirical background to the case studies

This section provides a background to the case stud-‐

ies, performed where indicated in Fig. 3.

Fig 3. Map of northern Europe. The grey surface comprises the two northernmost counties in Sweden and mainly for-‐

ested land. The black surface comprises Stockholm County and the largest metropolitan area in the Nordic countries.

Conservation and change in boreal forests (Q1) Forest ecosystem services are fundamental to the human communities in northern Sweden. Mature

Paper 2

Paper 3, 4 and 5 Paper 1

forests are concentrations of valuable timber as well as of services more dependent on biodiversity, like hunting and recreation. Forest stands outside pro-‐

tected areas (PAs) in the boreal forest belt represent 26% of the forested area globally, which is frag-‐

mented at several scales (Riitters et al. 2000). The forest landscape in northern Sweden is seriously fragmented with a steady decline in amount of old forests due to selective cutting before the 1950s and thereafter large-‐scale clear-‐cutting (Axelsson &

Östlund 2001; Esseen et al. 1997).

During the 20^th century PAs were increasingly set aside for wildlife and recreation, as the surrounding natural habitats were swept away by industrial for-‐

estry, agriculture and urbanization. Although area protection remains the most important measure to protect biodiversity, the focus has shifted from ”is-‐

land conservation” to systems of connected reserves, as awareness has grown about connectivity and resil-‐

ience, including the risks of genetic isolation, climate change and other consequences of altered distur-‐

bances regimes and renewal cycles (Bengtsson et al.

2003). Old pine trees, which constitute a habitat fac-‐

tor in paper 1 and 2, were favored by wildﬁre that was previously the dominating disturbance in boreal for-‐

ests, and favored to some extent also by intensive reindeer grazing (Hellberg et al. 2004), whereas they were removed by the selective cutting that dominated the subsequent regime in which ﬁre was suppressed (Granström 2001). While patch area and isolation af-‐

fect the size, frequency and intensity of disturbances like fire in forest patches (Baker 1989), alterations of a patch's disturbance regime may be determined to an even greater degree by the surroundings, which may transmit disturbance or modify the connectivity to other patches. For example, the probability of both fire ignition and rate of fire spread have been shown to differ across different types of human-‐modified habitat (Cochrane et al. 1999). In paper 1 and 2, our evaluation of the connectivity of patch configurations relates to patch-‐scale habitat qualities (paper 1) and different hypothetical interactions with the "matrix", including connectivity to stone fields (paper 1) and to the (harvest) disturbance regime outside the focal protected areas (paper 2).

(13)

In Scandinavian forests, there are presently too few large stands of old-‐growth forest to ensure a func-‐

tional and resilient reserve system (Bengtsson et al.

2003). As a response, conservation planning is look-‐

ing beyond the borders of PAs into the surrounding forests dominated by logging. Swedish policy during the last decades has sought to provide incentives for forest companies to harvest some areas less inten-‐

sively, however this has not yet succeeded and the Swedish Forest Agency (2011) found that every third felling in Sweden did not even fulﬁll the minimal le-‐

gal standards of biodiversity consideration. Still, there seems to be huge potential beneﬁts in a more integra-‐

tive spatial planning that more carefully takes into account the complementarity of areas with diﬀerenti-‐

ated prioritization of wood production and biodiver-‐

sity, including the use of dynamic (temporary) pro-‐

tection (Bengtsson et al. 2003). The many dimensions of forest protection has resulted in multifaceted de-‐

bates like the SLOSS debate (Single Large Or Several Small; Kingsland 2002); dynamic/temporary PAs; pro-‐

tection of spiritual forests; dispersal stepping-‐stones in agricultural landscapes and for maintaining soil and water quality. For example, around lake Bornsjön in Sweden, the water company Stockholm Vatten made an agreement with forest owners to spare large buﬀer zones as means to protecting the lake as a source of drinking water (Dudley & Phillips 2006).

Papers 1 and 2 probe Q1 by examining the response of species to the spatial complexity emerging from for-‐

est logging and protection. They explore the landscape-‐ecological mechanisms that aﬀect speciﬁc species (paper 1) and trait-‐based species groups (pa-‐

per 2). The papers hypothesize that species depend on protected forest with connectivity to geological stone ﬁelds (paper 1) and to other protected areas (paper 2). These two types of connectivity are mod-‐

eled more in detail using connectivity graphs taking into account connectivity in the “unprotected matrix”

as lower traversal cost (paper 1) or connector nodes (paper 2) (also see the methods chapter).

Wetlands in an urban region (Q3)

Since 1900 more than half of the world’s wetlands have been destroyed and in 1975 it was the ﬁrst major ecosystem to be protected by an international treaty (Zedler & Kercher 2005). Previous research has at-‐

tributed the failures of wetland protection policies to the complexity and "invisibility" of spatial relation-‐

ships among wetland water and vegetation (Turner et al. 2000). Following habitat loss per se, lacking con-‐

nectivity is the second largest threat to biodiversity in wetland systems (cf. Amezaga et al. 2002, Baldwin 2011).

Connectivity planning is challenging in urban land-‐

scapes where land-‐use intensiﬁcation and dispersal barriers signiﬁcantly constrain the dispersal of many species (Niemelä 2011). I approach Q3 in paper 5 by assessing the potential of current collaborations across municipal boundaries when it comes to man-‐

aging the connectivity of fragmented urban wetland systems. Urban wetlands are biodiversity hotspots and provide local communities with important eco-‐

system services like clean water, ﬂood control, recrea-‐

tion and supply of food and materials. However, ur-‐

banization has driven the ﬁlling, draining and dredg-‐

ing of most urban wetlands. Several studies on urban wetlands have found negative correlations between plant diversity and inter-‐wetland distances (e.g., Lopez & Fennessey 2002). Much of Sweden’s biodi-‐

versity is associated with wetlands, including more than half of the vascular plants and bird species, 40%

of the mosses and 70% of the land snails (CBM 2007).

In Stockholm County, 90% of the current wetland area has been aﬀected by peat mining and draining for farming and forestry (SCAB 2013). This has re-‐

sulted in severely fragmented wetland systems and reduced recolonization rates (Ministry of Environ-‐

ment 2012).

Paper 5 investigates Q3 by developing an integrative modeling approach of wetland connectivity coupled with a spatially distributed land-‐management system.

It evaluates the general degree as well as speciﬁc mismatches of ecological connectivity vs. collabora-‐

tive management across municipal boundaries. Our case study uses freely available GIS-‐data and visual-‐

izes the estimated mismatches and matches in the

(14)

landscape, facilitating interpretation and use of the results in local and regional management practice.

Municipal spatial planning and coordination (Q2+3)

Papers 3, 4 and 5 address Q2+3 in the context of the planning process in urban and semi urban munici-‐

palities in the Stockholm region. Relatively much green space remains within and around the urban cores, and the agency “Regional Growth, Environ-‐

ment and Planning” has identiﬁed ten green wedges that each extends over multiple municipalities (see ﬁgure 2 in paper 3 and SCAB 2010). Most municipali-‐

ties seek to take the green wedges into account in their planning, and municipal urban planners strug-‐

gle to accommodate an urban development while simultaneously protecting biodiversity and promot-‐

ing a sensible use of the region's natural resources.

Paper 3 studies how the municipalities assess connec-‐

tivity more locally, including how they may relate smaller landscape elements to the green wedges and the regional ecological complexity.

Municipal spatial planning is the governance process in Sweden with the largest inﬂuence on ecology out-‐

side the protected areas, which constitute only 5% of the land area in Stockholm County. The ”planning monopoly” in Sweden means that spatial planning is the exclusive responsibility of municipalities, and that the Plan-‐ and Building Acts since 1987 have con-‐

strained the interference from national authorities.

Municipalities must maintain a municipal-‐scale com-‐

prehensive (strategic) plan, which is implemented in block-‐scale detailed plans of speciﬁc projects, initi-‐

ated by the local government or by external develop-‐

ers. A few urban regions in Sweden have regional planning agencies, but regional plans have been po-‐

litically approved only in the Stockholm County.

Countries diﬀer with respect to how planning respon-‐

sibilities are assigned to local, regional and national governance levels, yet everywhere planning outcomes are inﬂuenced by the communication of ecological knowledge within planning agencies, and potentially

by coordination across diﬀerent jurisdictions. Paper 3 addresses Q2 by probing how the issue of ecological connectivity is considered in the planning process, and examines the pros and cons of using a CG soft-‐

ware tool, as perceived by municipal practitioners who provide ecological advice in the planning proc-‐

ess. Paper 4 widens the scope to ecological knowledge in general, and investigates the municipal ecologists' techniques and challenges when it comes to getting their knowledge across to planners and politicians.

Q3 is tackled by paper 5, estimating the potential of current collaboration between municipalities to deal with wetland connectivity across municipal borders.

Paper 3 and 5 also discuss whether coordination be-‐

tween regional and municipal plans may eﬀectively deal with ecological interactions across boundaries and scales.

Ecologists and knowledge in municipal planning practice (Q2)

Q2 is investigated in paper 3 and 4, looking into the communication of ecological knowledge in municipal administrations, which are required to implement diﬀerent national and European sustainability poli-‐

cies at the local level. Many municipalities employ ecologists who manage natural areas and who sup-‐

port the interpretation and operationalization of sus-‐

tainability policies in local practices like spatial-‐

planning. Limited use of ecological knowledge in planning is many times due to the diﬃculties in pre-‐

senting and interpreting information within planning processes (Yli-‐Pelkonen & Niemelä 2006), and to the failure of researchers to use a language apt to envi-‐

sion, negotiate, and manage the environmental con-‐

ditions of society (Norton 1998; Robertson & Hull 2001). As Roux et al. (2006) argue, planning is a knowledge domain that demand much synthesis and input from the domains of research (basic and ap-‐

plied), operational management and local communi-‐

ties; and this knowledge must be well codiﬁed (made explicit) for implementation in planning to be suc-‐

cessful. Hence, there is a need for knowledge brokers who may connect the knowledge of diverse domains through matching, sense-‐making and codiﬁcation/

Spatial complexity and fit between ecology and management: Making sense of patterns in fragmented landscapes

Spatial complexity and ﬁt between ecology and management Making sense of patterns in fragmented landscapes

Arvid Bergsten

Table of contents

1. Introduction 7

2. Background 10

3. Methods 15

4. Results and discussion of the papers 18

5. Future research 23

6. Conclusions for governance 26

Acknowledgements 26

References 27