Working PaperPDF Available

Identifying the Potential for Cross-Fishery Spillovers: A Network Analysis of Alaskan Permitting Patterns

Authors:

Abstract and Figures

Many fishermen own a portfolio of permits across multiple fisheries, creating an opportunity for fishing effort to adjust across fisheries and enabling impacts from a policy change in one fishery to spill over into other fisheries. In regions with a large and diverse number of permits and fisheries, joint-permitting can result in a complex system, making it difficult to understand the potential for cross-fishery substitution. In this study, we construct a network representation of permit ownership to characterize interconnectedness between Alaska commercial fisheries due tocross-fishery permitting. The Alaska fisheries network is highly connected, suggesting that most fisheries are vulnerable to cross-fishery spillovers from network shocks, such as changes to policies or fish stocks. We find that fisheries with similar geographic proximity are more likely to be a part of a highly connected cluster of susceptible fisheries. We use a case study to show that preexisting network statistics can be useful for identifying the potential scope of policy-induced spillovers. Our results demonstrate that network analysis can improve our understanding of the potential for policy-induced cross-fishery spillovers.
Content may be subject to copyright.
1616 P St. NW
Washington, DC 20036
202-328-5000 www.rff.org
November 2017 RFF WP 17-23
Identifying the
Potential for Cross-
Fishery Spillovers: A
Network Analysis of
Alaskan Permitting
Patterns
Ethan T. Addicott, Kailin Kroetz, Matthew Reimer,
James N. Sanchirico, Daniel K. Lew, and Justine
Huetteman
WORKING PAPER
© 2017 Resources for the Future. All rights reserved. No portion of this paper may be reproduced without
permission of the authors.
Resources for the Future (RFF) is an independent, nonpartisan organization that conducts rigorous economic
research and analysis to help leaders make better decisions and craft smarter policies about natural resources and the
environment.
Working papers are research materials circulated by their authors for purposes of information and discussion. They
have not necessarily undergone formal peer review. Unless otherwise stated, interpretations and conclusions in RFF
publications are those of the authors. RFF does not take institutional positions.
Identifying the Potential for Cross-Fishery Spillovers:
A Network Analysis of Alaskan Permitting Patterns
Ethan T. Addicott, Kailin Kroetz, Matthew Reimer,
James N. Sanchirico, Daniel K. Lew, and Justine Huetteman
Abstract
Many fishermen own a portfolio of permits across multiple fisheries, creating an opportunity for
fishing effort to adjust across fisheries and enabling impacts from a policy change in one fishery
to spill over into other fisheries. In regions with a large and diverse number of permits and
fisheries, joint-permitting can result in a complex system, making it difficult to understand the
potential for cross-fishery substitution. In this study, we construct a network representation of
permit ownership to characterize interconnectedness between Alaska commercial fisheries due to
cross-fishery permitting. The Alaska fisheries network is highly connected, suggesting that most
fisheries are vulnerable to cross-fishery spillovers from network shocks, such as changes to
policies or fish stocks. We find that fisheries with similar geographic proximity are more likely to
be a part of a highly connected cluster of susceptible fisheries. We use a case study to show that
preexisting network statistics can be useful for identifying the potential scope of policy-induced
spillovers. Our results demonstrate that network analysis can improve our understanding of the
potential for policy-induced cross-fishery spillovers.
Key Words: Fishing portfolios, leakage, network theory, spillover effects, fishery policy
JEL Classification Numbers: Q22
Contents
1. Introduction ......................................................................................................................... 1
2. Material and Methods ........................................................................................................ 3
2.1. Alaska Fisheries ........................................................................................................... 3
2.2. Data .............................................................................................................................. 4
2.3. Methods........................................................................................................................ 5
2.3.1. The Network .........................................................................................................6
2.3.2. Characterizing Network Topology .......................................................................7
2.3.3. Cluster Analysis ...................................................................................................9
2.3.4. Characterizing Spillover ....................................................................................11
3. Results and Discussion ...................................................................................................... 12
3.1. Network Statistics ...................................................................................................... 12
3.2. Cluster Analysis ......................................................................................................... 14
3.3. Spillover Case Study: ITQs in the Bristol Bay King Crab Pot Fishery ..................... 15
4. Conclusion ......................................................................................................................... 16
References .............................................................................................................................. 19
Tables and Figures ................................................................................................................ 22
Resources for the Future Addicott et al.
Identifying the Potential for Cross-Fishery Spillovers:
A Network Analysis of Alaskan Permitting Patterns
Ethan Addicott, Kailin Kroetz, Matthew Reimer,
James N. Sanchirico, Daniel K. Lew, and Justine Huetteman
1. Introduction
Most commercial fishery policies are directed towards a single species in a particular
region and caught by a certain gear, even though fishermen participate in multiple regions, catch
multiple species, and use multiple gears. Policy assessments that fail to account for the richness
of fishing enterprises can result in an incomplete understanding of the socioeconomic
consequences of management changes in target and non-target fisheries, as well as in fishing
communities. For example, fishermen may shift fishing effort to other fisheries in response to a
policy change in the target fishery. This so-called “leakage” can result in both positive and/or
negative spillover effects on both fishermen and communities involved in the impacted fisheries
(e.g., Cunningham et al. 2016). Furthermore, participation linkages across fisheries are especially
important to understand given recent work showing that a diversified portfolio of fisheries can
help fishermen and communities mitigate risk and smooth incomes (Kasperski and Holland
2013, Sethi et al. 2014a, Anderson et al. 2017, Cline et al. 2017). The mismatch between the
scope of past management policies and the reality of fisherman decision-making is an impetus
for expanding our understanding of multi-fishery participation.
Addicott: School of Forestry and Environmental Studies, Yale University, New Haven, CT 06460; email:
ethan.addicott@yale.edu. Kroetz: Resources for the Future, Washington, DC 20036; email: kroetz@rff.org. Reimer:
Institute of Social and Economic Research, Department of Economics and Public Policy, University of Alaska
Anchorage, 3211 Providence Drive Anchorage, AK 99508; email: mreimer2@alaska.edu. Sanchirico: Resources for
the Future, Washington, DC 20036; Department of Environmental Science and Policy, University of California,
Davis, California 95616; email: jsanchirico@ucdavis.edu. Lew: AFSC, NMFS, NOAA; Department of
Environmental Science and Policy, University of California, Davis, California 95616; email: dan.lew@noaa.gov.
Huetteman: Resources for the Future, Washington, DC 20036; email: huetteman@rff.org.
Disclosure statement: The authors are not aware of any affiliations, memberships, funding, or financial holdings that
might be perceived as affecting the objectivity of this review.
Acknowledgements: We thank Jacqueline Ho for research assistance. Funding for this research was provided by
Resources for the Future and the North Pacific Fisheries Research Board (NPRB). The views and opinions
expressed in this paper are the authors’ own, and do not necessarily reflect those of Yale University, Resources for
the Future, University of Alaska, the University of California, Davis, the National Oceanic and Atmospheric
Administration, or the U.S. Department of Commerce.
Resources for the Future Addicott et al.
The likelihood and degree of leakage from a single-fishery policy depends on the extent
to which fishermen can substitute between the target fishery and other fisheries. Ex ante
predictions of spillover impacts therefore require policy makers to identify fisheries that are
susceptible to leakage from a proposed single-fishery management action. To do so, policy
makers must first understand the prevailing patterns of cross-fishery participation. For instance,
to what extent are permits for the target fishery held in common with permits for other fisheries?
Are there fisheries that are more heavily connected to other fisheries through joint permitting,
making them more likely to generate spillover effects into other fisheries? Are there fishery
characteristics that make joint permitting and spillover impacts from a management change more
likely?
In this paper, we investigate the degree of interconnectedness and characteristics of
jointly permitted fisheries in the Alaska region to better understand the potential for single-
fishery policy spillovers. Despite the extensive work on the benefits of diversification of
individual and community fishing portfolios in Alaska (Kasperski and Holland 2013, Sethi et al.
2014a, Anderson et al. 2017, Cline et al. 2017), as well as analyses outside Alaska demonstrating
that single-fishery management actions can spill over into other fisheries (Asche et al. 2007,
Hutniczak 2014, Cunningham et al. 2016), there is limited information in Alaska on cross-fishery
permitting patterns and the scope for potential spillover effects from single-fishery policies.
We use tools and methods derived from network theory to characterize fisheries
interconnectedness in Alaska, and investigate how interconnectedness varies based on fishery
characteristics, including the areal extent of the fishery (i.e., statewide versus regional) and
management jurisdiction (i.e., federally-managed versus state-managed fisheries). As
demonstrated by Fuller et al. (2017), a particular strength of network statistics is that they can
capture fishery connectivity that extends beyond the first-degree, which allows us to examine the
potential for spillover impacts to spread beyond those fisheries directly connected to the target
fishery. We identify clusters of fisheries whose permits are more likely to be held in common
with other fisheries in the same cluster, but less likely with fisheries outside the cluster. This
allows us to examine the likely scope for potential spillover effects from single-fishery policies
and to identify characteristics of fisheries that tend to have a relatively high degree of cross-
fishery permitting. We consider several fishery characteristics that have either been shown or
hypothesized to be associated with cross-fishery permitting: using the same gear (Radtke and
Davis 2000), targeting the same or similar species (Radtke and Davis 2000, Cunningham et al.
2016), and/or geographic proximity (Sethi et al. 2014a). Finally, to examine the effectiveness of
network methods for identifying the potential for spillover, we explore whether the network of
Resources for the Future Addicott et al.
fisheries connected to the Bristol Bay King crab pot (BBK) fishery prior to the implementation
of the Individual Transferable Quota (ITQ) program in 2004 serves as a good proxy for changes
that took place after ITQ introduction.
Our results demonstrate that network analysis can greatly contribute to our understanding
of policy-induced cross-fishery spillovers. We find that the Alaska fisheries network is highly
connected--all but one of the fisheries form a connected graph, with approximately one-third of
all potential fishery pairs sharing at least one permit holder. Furthermore, we find that there is
significant heterogeneity in interconnectedness across fisheries, suggesting there is variability in
fisheries’ likelihood of generating and being susceptible to network shocks, such as changes to
policies or fish stocks. Fisheries that are federally managed and/or have a statewide geographic
scope tend to be larger and more likely to be core fisheries in the network—i.e., they share
permit holders with more fisheries, share a greater number of permit holders, and are more
central to the network, on average. We find that fisheries that share similar geography—and to a
lesser extent fishing gearare more likely to be a part of a highly connected cluster of fisheries,
suggesting that geographically similar fisheries may be a good place to look for potential
spillover effects from a management change in one of Alaska’s fisheries. Moreover, significant
cross-fishery permitting exists between federal and state fisheries, highlighting the importance of
considering potential impacts on state fisheries when designing and evaluating federal fishery
management policy, and vice versa. Finally, our case study of the BBK fishery indicates that
preexisting network statistics can be useful for identifying the potential scope of cross-fishery
policy-induced spillovers.
2. Material and Methods
2.1. Alaska Fisheries
There exist a variety of state- and federally-managed commercial fisheries in the waters
off Alaska, spanning multiple targeted species, gear types, locations, and management
institutions. Commercial fisheries that occur within three nautical miles from shore are managed
by the State of Alaska, while fisheries that occur greater than three nautical miles offshore are
federally managed.1 These Alaskan fisheries are prosecuted by both residents and non-residents,
1 There exists a subset of fisheries that occur more than three nautical miles offshore within the U.S. exclusive
economic zone for which management is delegated to the State (e.g., crab fisheries). For more detail on Alaska
fisheries see Appendix.
Resources for the Future Addicott et al.
using a variety of vessel types, ranging from small skiffs using longlines to catch Pacific halibut,
to large catcher processors, which catch and process pollock in the Bering Sea (NPFMC 2012).
The Alaska region is an important area to study with regard to cross-fishery participation
because (1) permit holders in Alaska fisheries often hold permits for more than one fishery (Sethi
et al. 2014b); (2) there is concern over how regulatory changes affecting fishery participation
impact Alaskan fishing communities (Carothers et al. 2010, Knapp 2011, Carothers 2013),
particularly because Alaskan fisheries are considered at high risk for catch and revenue
variability (Sethi et al. 2012); and (3) fishery portfolios in Alaska have become increasingly less
diverse over time, and as a result, variability of individual- and community-level fishing income
has increased (Kasperski and Holland 2013, Sethi et al. 2014a, Sethi et al. 2014b).
The existing literature on fishery participation in Alaska has documented individual and
community fishing permit and landing portfolio diversification over time. For example, Sethi et
al. (2014b) develop a metric-based approach to monitor the status of Alaska fishing communities
over time, and show that both the average number of permits held within a community and the
average number of permits held by a permit holder have declined over the past two decades.
Carothers et al. (2010) and Knapp (2011) show that there has been a disappearance of fishing
permits from rural Alaskan communities since the implementation of rights-based fishery
policies, such as limited-entry programs in the salmon fisheries and individual fishing quotas in
the halibut fishery. Other work has focused on the stabilizing role of fishery diversification on
fishing revenues. For example, Kasperski and Holland (2013) show that income stemming from
multiple fisheries in Alaska and Washington has become increasingly less diverse over time, and
as a result, variability of income from fishing has increased. Sethi et al. (2014a) provide evidence
that both fishery portfolio size and diversification play important roles in stabilizing community-
level fishing revenues.
2.2. Data
Any individual that participates in commercial fishing activity in State of Alaska waters,
including harvesting or landing catch from a state- or federally-managed fishery, requires a
fishery-specific permit issued by the Alaska Commercial Fisheries Entry Commission (CFEC).
CFEC permits are fishery-specific, with fisheries defined as area, gear, and species
combinations. In 1973, entry into a subset of Alaskan commercial fisheries was limited with the
goal of conserving Alaska’s fishery resources. A fixed number of permanent limited-entry
permits were issued. These permits must be renewed annually, regardless of whether the permit
holder chooses to participate in the fishery for which they are permitted. If the permit holder
Resources for the Future Addicott et al.
elects not to fish the permit or fails to renew it, the permit is removed from the fishery and not
reissued. At present, limited-entry permits cannot be leased except in medical emergencies;
however, they may be transferred or sold to other individuals by the permit holder. The CFEC
also issues interim-use permits to individuals wanting to access other non-limited Alaska
commercial fisheries or to limited-entry applicants awaiting qualification for permanent permits.
Interim-use permits are issued annually for non-limited entry commercial fisheries and are not
fixed in number.
Our analysis is based on the CFEC’s publicly available database of fishery permits
(available at: https://www.cfec.state.ak.us/fishery_statistics/permits.htm). Permit registration
information in the CFEC database includes both identifying information about the permit holder
and permit specific details including the fishery to which the permit grants entry, and the status
(current/former owner), type (permanent, interim, transferable), and ownership sequence of the
permit. In addition, we assign each fishery to a management jurisdiction (state or federal) and a
geographic extent (regional or statewide).2 We examine average permitting patterns over the past
10 years since the last major catch share program was implemented in Alaska (2006-2015) to
provide a recent representation of cross-fishery participation in Alaska.3 We also use data from
2004 and 2005 to investigate cross-fishery permit flows with the introduction of ITQs in the
BBK fishery.4
2.3. Methods
We use analytical and visualization methods derived from network theory, which is
conducive to studying relationships amongst discrete objects within complex systems. Similar
approaches that focus on understanding associations and connectivity within a complex system
have been applied to social-environmental systems (e.g., Dee et al. (2017)) and in multiple
contexts within the fisheries domain, including: information-sharing and social interactions
2 Note that for some CFEC fishery designations, particularly for the groundfish fisheries, a single CFEC permit can
be used to fish in both a federal and state fishery. For these fisheries, we designate a CFEC fishery as “federally
managed” if the fishery has a significant federal component to it, based on harvest specification tables for the
federally-managed groundfish fisheries (see https://alaskafisheries.noaa.gov/harvest-
specifications/field_harvest_spec_year/2016-2017-751). A more quantitive approach for assigning federal
management designations would require landings data, which are not publicly available.
3 Due to changes in CFEC fishery designations over the period of analysis, we group some fisheries to form a
consistent time series. There are 113 fisheries included in the 2006-2015 analysis. Fisheries with fewer than five
permit holders were dropped. See the Appendix for more details and for a list of all fisheries and their management
and geographic designations.
4 There are 114 fisheries and 116 fisheries included in the analysis in 2004 and 2005, respectively. Fisheries with
fewer than five permit holders were dropped.
Resources for the Future Addicott et al.
between fishery stakeholders (e.g., Barnes et al. (2016)), governance networks and regulatory
processes (e.g., Hartley (2010) and Mulvaney et al. (2015)), and international seafood trade
flows (e.g., Gephart and Pace (2015)). Network analyses in these settings are conducted to
illustrate properties of social, regulatory, or trade connectivity. We employ a similar conceptual
framework of fisheries connectivity to Bodin (2017) and Fuller et al. (2017) and similar network
theory methods and statistics to Fuller et al. (2017), but focus on whether connectivity between
Alaska’s fisheries identifies the potential for cross-fishery spillovers from single-fishery policies.
For calculations and figures, we use the free open-source graph visualization program Gephi.
2.3.1. The Network
A network can be represented as a graph consisting of a set of nodes (V) and edges (E),
defined formally as G = (V, E). In a binary (or unweighted) graph, each fishery is a node, where
the size of a node represents the number of permit holders in a fishery, and an edge between two
nodes is a binary measure indicating whether joint permitting between two fisheries exists or not.
In a weighted graph, each edge also has an associated value. We depict the Alaska fisheries
network as a weighted graph, where the weight of an edge represents the number of individuals
permitted in both of the adjoined fisheries.5
The traditional representation of cross-participation between fisheries—i.e., a cross-
tabulation matrix of fishery participation (Figure 1, Radtke and Davis (2000), and (NPFMC
2012))—has a network analog. In graph theory, the cross-tabulation matrix can be viewed as a
weighted adjacency matrix, associated with a weighted graph. For example, Fishery A and
Fishery B in Figure 1 share 10 permit holders, so we would represent cross-fishery permitting
between these two fisheries with an edge weight of 10 connecting the two nodes.
We model the Alaska fishery system using the average number of fishermen permitted in
a fishery between 2006 and 2015, and analyze the full network comprised of all Alaska fisheries,
along with the following four “sub-networks”: 1) a sub-network comprised of fisheries that are
primarily federally managed; 2) a sub-network comprised of fisheries that are managed by the
State of Alaska; 3) a sub-network comprised of fisheries with a statewide geographic extent; and
4) a sub-network comprised of fisheries with a regional (i.e., not statewide) geographic extent.
5 In traditional social network analysis (SNA), one might define a node as an individual and an edge as a friendship
or other type of connection between two individuals. In other social science work, nodes have represented
publications and edges have connected a publication with those it cites (e.g. Hopcroft et al. (2004)). Examples of
weighted networks include a network of colloborations between authors where the weights represent number of co-
authored publications and a network of air transportiation where the weights represent number of flights along
particular routes (Barrat et al. 2004).
Resources for the Future Addicott et al.
The purpose for examining state and federal fisheries as separate sub-networks is to
determine whether cross-fishery permitting may differ by management jurisdiction. Further,
network statistics can yield insight into cross-fishery permitting between each sub-network by
looking at the difference between the full network and the jurisdiction-specific networks. For
example, comparing measures of fishery connectedness between the full network and the
jurisdiction-specific networks provides information on spillover potential from federal to state
fisheries (and vice versa) after a single-fishery policy implementation. Understanding the
potential for spillover between state and federal fisheries is highly relevant from a policy
perspective since the federal and state governments are responsible for their own management
program design and evaluation, and do not necessarily consider impacts beyond their
jurisdiction. For example, federal management documents that do calculate cross tabulations
(e.g. NPFMC (2012)) only do so for federal fisheries.
Similarly, the purpose of considering regional and statewide fisheries as separate
networks is to determine the potential geographic scope of single fishery policy spillovers. For
instance, if some or all statewide fisheries are only permitted by a subset of fishermen from a
particular region, a change in a statewide fishery policy may not have significant impacts on
fishermen and fisheries in other regions. If, on the other hand, statewide fisheries are fished by
fishermen from a large set of regions, the potential geographic scope of impacts would be larger.
Finally, we construct networks of Alaska fishery interconnectedness before and after ITQ
introduction in the BBK fishery. We compare the cross-participation in these two static networks
and examine the flow of permits that occurred after ITQs were introduced. This case study
provides an example of how network methods and statistics can be used to examine the impact of
a management change on cross-participation, and examines the extent to which ex-ante network
statistics can identify the ex-post spillover from implementing a single-fishery management
change.
2.3.2. Characterizing Network Topology
We consider a range of commonly used indicators that include both aggregate measures
of an entire network, as well as node-specific measures of fishery connectedness in the network.
Aggregate network statistics provide information on characteristics of the entire graph. A basic
aggregate statistic is a binary measure of whether the graph is connected, meaning all fisheries
(nodes) are reachable via a path (i.e., a set of edges) from any other fishery. If the graph is not
connected, then there exist disconnected subgraphs, or portions of the network, that are
unconnected from other portions. The fisheries in a disconnected subgraph do not share permit
Resources for the Future Addicott et al.
holders with the subset of fisheries outside of the subgraph, and would therefore be unlikely to
experience spillover impacts from a policy change in a fishery outside the subgraph. The
smallest possible disconnected subgraph is an unconnected fishery, within which a permit holder
holds a permit for only that fishery.
Network density is an aggregate network statistic that measures the degree of network
connectivity, which in our case represents the extent to which fisheries share permit holders.
Network density is calculated by dividing the number of edges present in the network by the total
number of possible edges. The total number of possible edges is N×(N-1)/2, where N is the total
number of nodes. The actual number of edges equals the number of pairs of fisheries that share
at least one permit holder in one of the years 2006-2015. A network’s density can be interpreted
as the proportion of all fishery pairs that have at least one common permit holder; thus, a
network with low density is comprised of relatively few jointly-permitted fisheries, representing
a relatively unconnected network. Network density is useful for describing interconnectedness
because it controls for the number of fisheries in the network, which is important when
comparing connectivity among subsets of fisheries (e.g., statewide and regional fisheries).
However, density is limited because it does not consider the number of permit holders that are
jointly permitted in other fisheries.
Node-specific statistics are informative on their own and can also be aggregated to yield
insight into the entire network. We focus on node size, average edge weight, degree, and node
closeness centrality. Node size is calculated as the average number of permit holders in a fishery.
The average edge weight for a given node equals the average weight across all the edges
connected to the node (e.g., in Figure 1 the average edge weight of Fishery A equals 10; the
average edge weight of Fishery B equals 40. Average edge weight for a fishery therefore equals
the average number of permit holders a fishery shares with another fishery it is connected to. The
average edge weight across all edges in the network provides aggregate information on average
joint-permitting across all pairs of connected fisheries in the network.6
The degree of each fishery is the number of other fisheries that it is directly connected via
an edge. A node with a high degree represents a fishery that is jointly permitted with many other
fisheries. In contrast, a node with low degree represents a fishery comprised of fishermen
(potentially many) that rarely hold permits in other fisheries, and therefore displays little (if any)
joint permitting with other fisheries. The average degree of the network is the average of the
6 When calculating the aggregate average edge weight, pairs of fisheries that have no joint permit holders are not
included.
Resources for the Future Addicott et al.
degree across all network nodes. While node degree provides information on the extent of a
fishery’s first-degree connections, it may provide only limited information on the potential scope
of a policy spillover because it does not consider fishery connections beyond the first degree.
Node closeness centrality measures how connected a fishery is to all other fisheries in a
connected graph. The paths through which one fishery can be reached from other fisheries can
be thought of as a potential chain of spillover impacts. Calculation of node closeness centrality
relies on the distance, or path length, between two nodes, which is equal to the minimum number
of edges one must transverse to connect the two nodes. For example, if two fisheries share at
least one permit holder, such as fisheries A and B in Figure 1, the distance between the two
nodes is one, whereas the distance between fisheries A and C is equal to two. This measure of
connectedness is an extension to basic cross tabulations because it goes beyond first-degree
connections, thereby capturing the potential for spillover impacts to spread beyond directly-
connected fisheries through a chain of spillover impacts. Letting d(vi,vj) denote the distance
between nodes i and j, the closeness centrality of a node i is given by:
 =
,

, ,
which is the inverse of the average distance between a node and all other nodes. The
closeness centrality is between zero and one. If a fishery is directly connected to every other
fishery via an edge, its closeness centrality will be one; on the other hand, if a fishery shares
relatively few edges with other fisheries, the distance between nodes will be higher and the
closeness centrality will gravitate towards zero. For example, if a fishery shares permit holders
with many other fisheries, and in addition, those permit holders also share connections with the
remaining fisheries, then that fishery would have a large measure of closeness centrality. We
calculate average node centrality over the network by averaging the node centrality across all
nodes in the network. This provides information at the network level on how connected fisheries
are to other fisheries, considering both direct connectivity (two fisheries sharing permit holders)
as well as indirect connectivity (the extent to which a fishery shares permit holder with fisheries
that also share permit holders with other fisheries, and so on).
2.3.3. Cluster Analysis
We use cluster analysis to partition the network into clusters of fisheries that exhibit high
levels of cross-fishery permitting within the cluster, but relatively low cross-fishery permitting
with fisheries outside the cluster. A fishery cluster therefore represents a set of fisheries from
Resources for the Future Addicott et al.
which fishermen choose when making permitting decisions, and thus likely represents historical
fishery substitution possibilities. Intuitively, a fishery cluster represents the likely scope for
potential spillover effects from a single-fishery policy.
We identify clusters using a modularity-based community detection algorithm to
statistically partition the network into clusters of fisheries (Blondel et al. 2008).7 We then
analyze clusters in terms of node and aggregate network characteristics, such as connectivity and
node centrality within the cluster, and the extent to which fishery characteristics are associated
with cluster membership.8 Specifically, we examine the extent to which cluster membership is
more common between fisheries that are geographically close, are fished using the same gear,
have a similar target species, and/or fall under federal or state jurisdiction. Fishery characteristics
that are good predictors of cluster membership aid in the identification of fisheries that may or
may not be susceptible to spillover effects from a single fishery policy.
To analyze cluster membership in terms of fishery characteristics, we estimate the
probability that two fisheries belong to the same cluster as a function of sharing fishery
characteristics. Specifically, for each potential fishery pair in the network, we calculate a binary
variable cluster equal to one if the fisheries are in the same cluster, and zero if not.9 Additionally,
we create separate binary variables indicating whether each pair of fisheries share a common
fishing gear, geographic area, and target species. We then estimate the probability that the
fisheries in pair i belong to the same cluster as a function of sharing the same gear, geographic
area, and/or target species using the following logistic regression model:
(= 1)=(+++),
7 Clusters are groups of nodes in a network that are characterized by a high density of internal connections and low
density of external connections. Modularity, Q, is the optimand for the Louvain cluster detection algorithm we
employ. It is determined by the ratio of the number of edges within clusters to the number of edges between clusters
across a network (Newman 2004). The number of clusters and assignment of nodes to clusters are chosen to
maximize network modularity. This partitions the network into clusters of nodes with higher edge representation
than had edges been assigned to the graph randomly. Cluster identification is used in a variety of fields to determine
functional groups to simplify complex networks for decision-making (Danon et al. 2005).
8 Network and node statistics parallel those presented for the aggregate network, but are calculated for only those
fisheries within a cluster.
9 Note that we combine CFEC species, gear, and area codes into more aggregated groups to facilitate the analysis of
clustering patterns. After dropping fisheries where a gear, area, or species attribute is “other”, we end up with 99
fisheries, which results in 4,851 fishery pairs (see Appendix for additional detail). We treat the statewide designation
as its own area code.
Resources for the Future Addicott et al.
where ()=(1 + ) is the cumulative logistic distribution. We then calculate each
characteristic’s independent influence on cluster membership as the average marginal effect of
sharing a characteristic on the probability that two fisheries belong to the same cluster. For
example, fishing gear’s influence on cluster membership is calculated as:

(= 1| = 1) 
(= 1| = 0),
where 
denotes the average predicted probability from the logistic regression.
2.3.4. Characterizing Spillover
We use the introduction of ITQs in the BBK fishery in 2005 as a means of exploring how
network methodology can be used to characterize joint permitting pre- and post-management
change, as well as examine spillover following a policy implementation.10 We focus on spillover
along the extensive margin—i.e., exit and entry decisions—because: 1) exit and entry are
relatively well-studied decisions and documented in the fishery management literature (e.g.,
Brinson and Thunberg (2016)); and 2) our data does not allow us to measure behavioral changes
along the intensive margin, such as changes in fishing effort in non-target fisheries.11 We
identify five types of permit-holder behavior coinciding with the ITQ (Figure 5): permit holders
who remain permitted in the BBK fishery; permit holders who exit the BBK fishery and who
have no other permits in other Alaska fisheries (these permit holders are determined to exit the
region); permit holders who exit the BBK fishery and do not change their fishing portfolio (i.e.,
do not obtain permits in new fisheries); permit holders who exit the BBK fishery and change
their fishing portfolio by obtaining permits in new fisheries; and permit holders who enter the
BBK fishery. Focusing on the permit holders who exit the BBK fishery and obtain permits in
new fisheries, we determine the proportion of these new fisheries that were in the preexisting
BBK network and compute the correlation between pre-ITQ edge weight and number of permit
acquisitions for those fisheries in the preexisting BBK network.
10 For more information on the fishery and post-management changes see (Fina 2005) and Reimer et al. (2014).
11 In general, there are several behavioral margins across which fishers can respond to a policy change (Smith 2012,
Reimer et al. 2014).
Resources for the Future Addicott et al.
3. Results and Discussion
3.1. Network Statistics
Table 1 (row 1) presents statistics for the full network of Alaska fisheries and Figure 2
provides a visual representation of the network from 2006-2015, where the size of a node scales
with the average number of permit holders in a fishery and the thickness of an edge represents
the average number of fishermen jointly permitted in the two fisheries.12 Both indicate that
Alaska fisheries are very connected through the cross-fishery permitting behavior of fishermen.
Indeed, all but one of the 113 fisheries form a connected graph and are therefore able to be
reached via a path from all other fisheries.13 Network density indicates that approximately one-
third of all possible edges are present, implying that one in three fisheries pairs are directly
connected through joint permitting, on average. Network degree indicates that the average
fishery shares permit holders with approximately 35 other fisheries, while average edge weight
indicates that a connected pair of fisheries share nearly 6 permit holders, on average. Altogether,
these statistics suggest that there is considerable potential for spillovers across fisheries in the
Alaska region.
The statistics in Table 2 indicate that larger fisheries (i.e., higher node size) that share
relatively more permit holders with other fisheries (i.e., more edges, higher edge weight) and
have higher closeness centrality often appear at the core, or center, of the network. For example,
the statewide halibut longline fishery (B06B), which appears at the center of Figure 2, is the
largest fishery in terms of number of permit holders, has the highest closeness centrality, and has
the highest degree in the full Alaska fisheries network. In contrast, fisheries with lower node-
level statistics tend to be located on the periphery of the network and not directly connected to
the core fisheries. For example, the scallop dredge fishery (W22BV), which has only 6 permit
holders and an extremely low closeness centrality (0.39) and average edge weight (less than 1),
is not connected to any of the core fisheries listed in Table 2.
The large measures of centrality common to many of the core fisheries are primarily
driven by two mechanisms, which together have important spillover implications. First, the
central fisheries have many direct, or first-order, connections with other fisheries through joint
permitting, as measured by node degree. Indeed, the four fisheries with the highest closeness
12 The layout of the fisheries in Figure 2 is based on an algorithm that takes into account the presence and weight of
edge relationships in order to represent relative connectivity of nodes. Node characteristics, such as the number of
permit holders and fishery characteristics, are not considered, though node arrangement may reveal trends about the
characteristics of fisheries that are more central to the aggregate network.
13 The single disconnected fishery in the full network is the Bering Sea Korean hair crab pot gear fishery (E91QV).
Resources for the Future Addicott et al.
centrality are also the four nodes with the highest degree. Second, many of the central fisheries
are directly connected to one another (Figure 2), creating indirect connections between fisheries
that do not directly share permit holders. Together, the direct and indirect connectivity creates a
network in which spillover from a single-fishery policy in a core fishery could impact both other
core and periphery fisheries. Further, given that many of the most central fisheries are also some
of the largest and have the highest edge weights, both the scope and the magnitude of single-
fishery policy spillovers from core fisheries have the potential to be large.
Network statistics for the four different sub-networks we consider (Table 1, rows 2-5)
indicate that considerable heterogeneity exists between the sub-networks. Indeed, many
federally-managed and statewide fisheries appear at the core of the network and differ in their
connectivity relative to state-managed and regional fisheries.14 The statewide and federally-
managed fisheries each a density, edge-weight, and closeness centrality that are considerably
larger than those of the regional and state-managed networks, suggesting that single-fishery
policies in federal and statewide fisheries may be likely to generate substantial direct spillover
across fisheries within their own respective networks. However, the substantial joint-permitting
across the four sub-networks suggests that spillovers are not likely to be contained within a
single sub-network (Table 3). Out of the 1,994 edges that comprise the entire network, 773 edges
(~39%) connect statewide fisheries to regional fisheries. Further, the 2,002 possible edges
between statewide and regional fisheries implies that 39% of all possible regional-statewide
fishery pairs are connected through joint permitting. Edges connecting state-managed to
federally-managed fisheries also make up a significant portion of the total number of edges in the
full network (453 out of 1,994 edges, or 23%) and have large average edge weight—on average,
there are 10.3 permit holders per federal-state fishery pair. Further, 38% of all possible state-
federal fishery pairs are connected through joint permitting. Altogether, these findings suggest
that single-fishery policies in federal and/or statewide fisheries have the potential to spill over
into state-managed and/or regional fisheries (and vice versa). Therefore, spillover impacts on
fisheries under other jurisdictions and in other regions may need to be considered in order to
estimate the full impact of a policy change.
14 Note that although many federal fisheries cover statewide areal extents, there is not a one-to-one correspondence
between these two designations. See Appendix for network figures of each of the four jurisdiction- and geographic-
specific networks.
Resources for the Future Addicott et al.
3.2. Cluster Analysis
The modularity-based community detection algorithm identified seven clusters of
fisheries that are relatively disconnected from fisheries in the other clusters: five clusters that
contain between 5 and 37 fisheries, and two clusters that contain only one fishery each (Table
4).15 Cluster 1 is the largest cluster in terms of the number of fisheries (37), is the most
internally connected, as measured by its density, average node degree, and average centrality,
and contains some of the most central fisheries in the full network—e.g., the federally managed
statewide halibut and sablefish longline fisheries, the federally managed longline groundfish
fisheries, the state-managed salmon troll fishery, and the southeast king crab pot fisheries.
Cluster 2 contains a significant number of pot gear fisheries targeting groundfish and crab—e.g.,
the federally managed groundfish pot fishery, the statewide groundfish jig fishery, and the BBK
fishery. The fisheries that comprise Clusters 3 and 4 are geographically grouped, with Cluster 3
being dominated by fisheries in southcentral and southeast Alaska—e.g., Prince William Sound
herring and shrimp pot fisheries and the Cook Inlet setnet salmon fishery—and Cluster 4 being
primarily comprised of fisheries in west and northwest Alaska—e.g., Bristol Bay salmon
fisheries, salmon and herring fisheries in Kuskokwim Bay and Norton Sound. Cluster 5 consists
of only dive fisheries, which do not appear in any other cluster, while Clusters 6 and 7 are single-
fishery clusters--the Bering Sea Korean hair crab pot fishery and the vessel-permitted scallop
dredge fishery.
Using the output from the logistic regression model, we reject the null hypotheses that
sharing a geographic area (p-value = 0.000) and sharing a fishing gear (p-value = 0.000) do not
have a marginal effect on the probability of being a member of the same cluster (Table 5). Of the
fishery pairs that share the same area (gear, species), 55% (34%, 23%) of fishery pairs are
predicted to be in the same cluster, whereas among the fishery pairs that do not share an area
(gear, species), only 16% (23%, 25%) are predicted to be in the same cluster. Thus, sharing the
same geographic area increases the probability of being in the same cluster by 0.387 percentage
points (from .162 to .549), which is considerably more than sharing a species (-0.0219) or gear
type (0.108). This finding can be seen visually as well in Figure 3, where we map fisheries based
on geographic location and cluster membership. Thus, geographic proximity is likely a natural
15 We find that several clustering variants fall within the tolerance of Gephi's optimization routine. We rerun the
optimization program and find the clusters are very robust (generally less than 5% of fisheries, which generally have
low centrality, move clusters) with one exception. In cases with fewer clusters the dive fisheries sometimes appear
in Cluster 1 or as split apart in Clusters 1 and 2. This pattern holds if we increase the resolution, thus decreasing the
number of clusters. For our analysis we keep the dive fisheries as their own cluster.
Resources for the Future Addicott et al.
starting point for identifying fisheries that could be susceptible to a single-fishery policy, a result
that is consistent with previous work indicating that a fishing community’s fishery portfolio may
be predetermined by the set of local fishing opportunities (Sethi et al. 2014a).
3.3. Spillover Case Study: ITQs in the Bristol Bay King Crab Pot Fishery
In Figure 4 we present the network for the BBK fishery in 2004 prior to the introduction
of ITQs (panel (a)) and the change in permits between 2004 and 2005 with the onset of ITQs
(panel (b)). The most notable change was the shrinking of the BBK fishery node size (labeled
K09T), reflecting the decrease in permit holders from 266 to 136 between 2004 and 2005. Of the
136 permit holders in 2005, 124 were permitted in the fishery prior to ITQs and 12 were new
entrants, implying that 142 permit holders exited the fishery after ITQs (Figure 5). Of those
permit holders that exited the fishery, 35 did not hold permits in any Alaska fishery in the
following year, 66 did not change their permit holdings in other fisheries, and 41 acquired a total
of 62 permits in other fisheries. The large amount of exiting resulted in a considerable drop in
cross-participation between the BBK fishery and other fisheries after ITQs. However, while the
average edge weight of the fishery decreased (from 12.4 in 2004 to 8.3 in 2005), there was not a
significant change in the number of connected fisheries (degree is 42 in 2004 and 40 in 2005) or
its overall connectivity to the network (closeness centrality is 0.60 in 2004 and 0.59 in 2005).
Thus, exit from the BBK fishery was relatively proportional to the joint-permitting pattern that
existed prior to ITQ introduction.
To explore whether ITQs in the BBK fishery generated spillover into non-target fisheries,
we focus on the 41 permit holders that exited the BBK fishery and acquired permits in new
fisheries.16 This set of permit holders obtained 62 new permits in 21 fisheries (Figure 4 panel
(b)), ranging from 13 permits in the Bering Sea Tanner crab pot fishery (T09Q) to 1 permit in
several other fisheries. While determining the magnitude of impacts in non-target fisheries from
spillover is beyond the scope of this paper, any impact likely depends on the characteristics of
the non-target fisheries, such as the number of permit holders. For example, the addition of 7
permit holders in the Statewide miscellaneous saltwater finfish longline (M06B) fisheries
16 In focusing on this set of permits, we explore permitting changes that coincided with the onset of ITQs. We leave
for future work a full causal estimation of the spillover, which requires identifying the counterfactual spillover that
would have occurred if the management change did not occur. The relatively high magnitude of new permit
acquisitions that coincides with the introduction of ITQs in the BBK fishery, however, suggests that much of this
respoonse is due to the ITQ program, as opposed to a simultaneous shock or individual motivations on the part of
the permit holder. Information on prior entry and exit of permit holders from the BBK fishery is available in the
Appendix.
Resources for the Future Addicott et al.
represents only a 1% increase in the number of permit holders in these fisheries, and therefore
did not likely have much of an impact on preexisting permit holders. In contrast, the Bering Sea
King crab pot fishery (K09Q) gained two permit holders from the BBK fishery, bringing its total
number of permit holders in 2005 to ten, an increase of 25%.
One of the potential benefits of network methodology is its capability of predicting the
scope of spillover from a policy change. As an example, to what extent is the pre-ITQ network
(Figure 4 panel (a)) able to predict the flow of permit holders after ITQ introduction in the BBK
fishery (Figure 4 panel (b))? Much of the change in permit holding coinciding with the ITQ
program is well-correlated with the pre-ITQ network structure and statistics. Prior cross-fishery
participation is a good predictor of where exiters of the BBK crab fishery enter. Of the 21
fisheries receiving new permit holders, 18 shared permit holders with the BBK fishery in 2004.
Only 2% of fisheries that did not share a prior connection with the target fishery received exiters,
while 40% of fisheries that did share a prior connection with the target fishery received exiters.
There is also evidence that the number of shared permit holders with the BBK fishery is a good
predictor of spillover; indeed, a positive correlation of 0.8 exists between the edge weight for
fisheries connected to the BBK fishery and the number of new permit acquisitions. For example,
pre-ITQ, the BBK fishery shared the highest edge weight with the Bering Sea Tanner crab pot
fishery (T09Q), which also received the highest number of exiters (13) from BBK post-ITQ.
Overall, this example demonstrates that network statistics, such as the degree and edge weights
associated with a fishery undergoing a policy change, can be useful for identifying the potential
scope for cross-fishery spillover.
4. Conclusion
Representing a complex system of fisheries as a network leads to a novel characterization
of cross-fishery permitting that has the potential to inform policy design and to identify fisheries
that are susceptible to spillovers from single-fishery policies. Our results demonstrate that cross-
fishery permitting in Alaska is substantial, spanning multiple management jurisdictions (federal
and state), species, geographic areas, and gear types. Several large federal and statewide fisheries
are central to the network, providing a means for nearly all fisheries in the network to be
connected to one another, either directly through joint permitting, or indirectly through
connected paths of jointly-permitted fisheries. On one hand, the significant cross-fishery
permitting makes it likely that management changes in one fishery could have far reaching
spillover effects on other fisheries and the fishermen and communities associated with these
Resources for the Future Addicott et al.
fisheries. On the other hand, there is heterogeneity in cross-fishery permitting that likely makes
some fisheries more susceptible than others to policy changes in other fisheries.
Several insights regarding cross-fishery permitting emerge from our analysis. First,
federal and statewide fisheries generally have the highest levels of cross-fishery permitting and
are the most central fisheries in the network, making them the most likely candidates for
generating spillover from single-fishery policies. Second, we find significant cross-fishery
permitting between federal and state fisheries, which highlights the importance of considering
potential impacts on state fisheries when designing and evaluating federal fishery management
policy, and vice versa. Third, clusters of well-connected fisheries tend to be composed of
fisheries within adjacent geographic areas, suggesting that geographically similar fisheries may
be a good place to look for potential spillover effects from a management change. Finally,
preexisting network structure and statistics can be useful for identifying the potential scope of
policy-induced spillovers, as demonstrated by our investigation of the spillover coinciding with
the ITQ program on the BBK fishery.
From a policy perspective, establishing baseline information on cross-fishery permitting
using network theory provides useful information to fishery managers and stakeholders in
developing a data-driven scope of fisheries and permit holders that are likely to be impacted by
proposed polices. However, the approach we present in this paper is limited in several respects,
and work remains before spillover effects from past and future single-fishery policies can be
fully evaluated and predicted. One area of future work concerns the fact that the data used in our
analysis contains information solely on permitting behavior, not fishing behavior (which is
confidential). In fact, several permits go unfished in any given year—a phenomenon often
known as “permit latency” (CFEC 2015). Therefore, spillover effects of single-fishery policies
would be better understood by examining permitting behavior and fishing effort (e.g., landings,
days fished) jointly. Another challenge is that retrospective evaluations of spillover effects from
previous policies require both a time series of network statistics and a strategy for isolating
policy-induced effects from other shocks that affect the evolution of the network. Other shocks
that could impact fishermen behavior include stock changes, that in turn could result in changes
to fishing season length and timing, which could influence cross-fishery permitting and
participation (Radtke and Davis 2000). Methods for estimating network statistics over time and
for causal policy evaluations have yet to be applied within a fisheries context. Lastly, permitting
behavior in the past is influenced by the prevailing institutions governing fisheries at that time.
For instance, many of Alaska state fisheries are governed by limited-entry programs, which
likely have the effect of reducing joint permitting to levels below those that would exist if those
Resources for the Future Addicott et al.
fisheries were open access. If a policy change influences the rules governing cross-fishery
participation, then future permitting behavior will not necessarily be the same as it was in the
past. In such a case, predicting the potential spillover effects from a single-fishery policy will
require a more mechanistic approach that is able to predict how fishermen make permitting
decisions and how these choices will change in response to a new policy.
Although more work is required to fully operationalize the potential for network methods
to inform fisheries policy, a network approach has the potential to contribute to efforts evaluating
socioeconomic impacts of ecosystem-based fisheries management (EBFM). EBFM programs are
increasingly being adopted around the world yet the magnitude and sign of social and economic
impacts of EBFM are unknown (Marshall et al 2017). Prior work has shown that aggregate
revenues can be increased and their variance decreased by taking an ecosystem-based approach
when setting catch levels (Sanchirico et al. 2008); a network-based approach to characterizing
cross-fishery permitting could improve our understanding of how changing catch levels could
impact fishermen.
Resources for the Future Addicott et al.
References
Anderson S. C., E. J. Ward, A. O. Shelton, M. D. Adkison, A. H. Beaudreau, R. E. Brenner, A.
C. Haynie, J. C. Shriver, J. T. Watson and B. C. Williams. 2017. Benefits and Risks of
Diversification for Individual Fishers. Proceedings of the National Academy of Sciences
114(40): 10797-10802.
Asche F., D. V. Gordon, C. L. Jensen, S. U. Rashid, G. R. Munro and J. G. Sutinen. 2007.
Individual Vessel Quotas and Increased Fishing Pressure on Unregulated Species. Land
Economics 83(1): 41-49.
Barnes M. L., J. Lynham, K. Kalberg and P. Leung. 2016. Social Networks and Environmental
Outcomes. Proceedings of the National Academy of Sciences 113(23): 6466-6471.
Barrat A., M. Barthelemy, R. Pastor-Satorras and A. Vespignani. 2004. The Architecture of
Complex Weighted Networks. Proceedings of the National Academy of Sciences of the
United States of America 101(11): 3747-3752.
Blondel V. D., J. L. Guillaume, R. Lambiotte and E. Lefebvre. 2008. Fast Unfolding of
Communities in Large Networks. Journal of Statistical Mechanics-Theory and
Experiment.
Bodin Ö. 2017. Collaborative Environmental Governance: Achieving Collective Action in
Social-Ecological Systems. Science 357(6352).
Brinson A. A. and E. M. Thunberg. 2016. Performance of Federally Managed Catch Share
Fisheries in the United States. Fisheries Research 179(Supplement C): 213-223.
Carothers C. 2013. A Survey of Us Halibut IFQ Holders: Market Participation, Attitudes, and
Impacts. Marine Policy 38(C): 515-522.
Carothers C., D. K. Lew and J. Sepez. 2010. Fishing Rights and Small Communities: Alaska
Halibut IFQ Transfer Patterns. Ocean & Coastal Management 53(9): 518-523.
CFEC. 2015. CFEC Permit Holdings and Estimates of Gross Earnings in the Bristol Bay
Commercial Salmon Fisheries, 1975-2014. R. Section.
Cline T. J., D. E. Schindler and R. Hilborn. 2017. Fisheries Portfolio Diversification and
Turnover Buffer Alaskan Fishing Communities from Abrupt Resource and Market
Changes. Nature Communications 8: 14042.
Cunningham S., L. S. Bennear and M. D. Smith. 2016. Spillovers in Regional Fisheries
Management: Do Catch Shares Cause Leakage? Land Economics 92(2): 344-362.
Resources for the Future Addicott et al.
Danon L., A. Diaz-Guilera, J. Duch and A. Arenas. 2005. Comparing Community Structure
Identification. Journal of Statistical Mechanics: Theory and Experiment 2005(09):
P09008.
Dee L. E., S. Allesina, A. Bonn, A. Eklof, S. D. Gaines, J. Hines, U. Jacob, E. Mcdonald-
Madden, H. Possingham, M. Schroter and R. M. Thompson. 2017. Operationalizing
Network Theory for Ecosystem Service Assessments. Trends in Ecology & Evolution
32(2): 118-130.
Fina M. 2005. Rationalization of the Bering Sea and Aleutian Islands Crab Fisheries. Marine
Policy 29(4): 311-322.
Fuller E. C., J. F. Samhouri, J. S. Stoll, S. A. Levin and J. R. Watson. 2017. Characterizing
Fisheries Connectivity in Marine Social–Ecological Systems. ICES Journal of Marine
Science 74(8): 2087-2096.
Gephart J. A. and M. L. Pace. 2015. Structure and Evolution of the Global Seafood Trade
Network. Environmental Research Letters 10(12): 125014.
Hartley T. W. 2010. Fishery Management as a Governance Network: Examples from the Gulf of
Maine and the Potential for Communication Network Analysis Research in Fisheries.
Marine Policy 34(5): 1060-1067.
Hopcroft J., O. Khan, B. Kulis and B. Selman. 2004. Tracking Evolving Communities in Large
Linked Networks. Proceedings of the National Academy of Sciences 101(suppl 1): 5249-
5253.
Hutniczak B. 2014. Increasing Pressure on Unregulated Species Due to Changes in Individual
Vessel Quotas: An Empirical Application to Trawler Fishing in the Baltic Sea. Marine
Resource Economics 29(3): 201-217.
Kasperski S. and D. S. Holland. 2013. Income Diversification and Risk for Fishermen.
Proceedings of the National Academy of Sciences 110(6): 2076-2081.
Knapp G. 2011. Local Permit Ownership in Alaska Salmon Fisheries. Marine Policy 35(5): 658-
666.
Mulvaney K. K., S. Lee, T. O. Höök and L. S. Prokopy. 2015. Casting a Net to Better
Understand Fisheries Management: An Affiliation Network Analysis of the Great Lakes
Fishery Commission. Marine Policy 57: 120-131.
Resources for the Future Addicott et al.
Newman M. E. 2004. Fast Algorithm for Detecting Community Structure in Networks. Physical
review E 69(6): 066133.
NPFMC. 2012. Fishing Fleet Profiles. North Pacific Fishery Management Council. 2017.
Radtke H. D. and S. W. Davis. 2000. Description of U.S. West Coast Commercial Fishing Fleet
and Seafood Processors. PSMFC.
Reimer M. N., J. K. Abbott and J. E. Wilen. 2014. Unraveling the Multiple Margins of Rent
Generation from Individual Transferable Quotas. Land Economics 90(3): 538-559.
Sanchirico J. N., M. D. Smith and D. W. Lipton. 2008. An Empirical Approach to Ecosystem-
Based Fishery Management. Ecological Economics 64(3): 586-596.
Sethi S. A., M. Dalton and R. Hilborn. 2012. Quantitative Risk Measures Applied to Alaskan
Commercial Fisheries. Canadian Journal of Fisheries and Aquatic Sciences 69(3): 487-
498.
Sethi S. A., M. N. Reimer and G. Knapp. 2014a. Alaskan Fishing Community Revenues and the
Stabilizing Role of Fishing Portfolios. Marine Policy 48(C): 134-141.
Sethi S. A., W. Riggs and G. Knapp. 2014b. Metrics to Monitor the Status of Fishing
Communities: An Alaska State of the State Retrospective 1980-2010. Ocean & Coastal
Management 88(c): 21-30.
Smith M. D. 2012. The New Fisheries Economics: Incentives across Many Margins. Annual
Review of Resource Economics 4(1): 379-402.
Resources for the Future Addicott et al.
Tables and Figures
Figure 1: Cross-fishery permitting example: number of individuals permitted in each pair
of fisheries. The adjacency matrix in Panel a is symmetric; diagonal values are total individuals
permitted in each fishery; off-diagonals must be strictly less than row and column diagonal
numbers (i.e., cross-permitting is less than or equal to the minimum number of individuals in the
two fisheries). Panel b represents the same adjacency matrix as a network diagram.
(a)
Fishery A
Fishery B
Fishery C
Fishery D
Fishery A
100
10
0
0
Fishery B
-
110
70
40
Fishery C
-
-
200
35
Fishery D
-
-
-
50
(b)
Resources for the Future Addicott et al.
Figure 2: Cross-fishery permitting in the Alaska fishery network, 2006-2015. Nodes represent
individual fisheries, with node size corresponding to the average number of permits held for that
fishery. Edge thickness represents the average number of permit holders permitted in each of the
two connected fisheries.
Note 1: Two nodes are excluded from this figure. One, E91QV, is not connected to the network.
The second, W22BV, was plotted by the algorithm far away from the network but connected to
three nodes. Both were removed to maintain graph clarity.
Note 2: As described in the text and the Appendix, in some instances we aggregated across CFEC
codes to form a fishery unit for the analysis. These instances are shown in the table with the
species, gear, and area designation used in Figure 2 to label the node listed first, followed by the
codes that were aggregated to form the fishery unit for the analysis.
Resources for the Future Addicott et al.
Resources for the Future Addicott et al.
Figure 3: Geographic Location of Area-Designated Fisheries. Node size represents relative
number of permits issued from 2006-2015 in each fishery. The colors represent the clusters from
the cluster analysis. Nodes are placed in the approximate location of the associated fishable area.
Note: Only clusters 1-5 appear in the figure.
Resources for the Future Addicott et al. 2017
1
Figure 4: Bristol Bay King Crab Fishery 2004-5. Node size represents relative number of
permits issued for the fishery. Edge weight represents number of permit holders jointly permitted
in those two fisheries.
Note: Only nodes with five or more permit holders are shown. The edges between BBK and its
connected fisheries have been magnified in order to facilitate visual analysis and thus do not scale
to the gray edges not connected to BBK. 2004 (a)
Spillover (b)
Resources for the Future Addicott et al. 2017
Figure 5: Flow of Permits in the Bristol Bay King Crab Pot Fishery (BBK), 2004-5: Following
the implementation of the ITQ program, the following actions were taken by permit holders in the
BBK fishery.
Note: The number of individual permit holders taking the action is in parentheses.
Resources for the Future Addicott et al. 2017
Table 1: Network Summary Statistics.
Total
Number of
Fisheries
Fisheries
Connected
Fisheries
Unconnected
Density =
Edges/
Potential
Edges
Average
Node Size
Average
Edge Weight
Average
Degree
Average
Closeness
Centrality
Entire Network
113 112 1 .32
177
5.82 35.29 .59
Fishing Region
Statewide Only
22
22
0
.66
311
17.67
13.91
.77
Regional Only
91 90 1 .26
145
3.70 23.47 .55
Management Entity
Federal Managed
Only
10 10 0 .98
406
31.16 8.8 .98
State Managed Only
102 101 1 .29
156
3.72 29.35 .57
`
Resources for the Future Addicott et al. 2017
Table 2: Alaska Fishery (Node) Characteristics. Characteristics of fisheries for the statewide
and regional fisheries with the highest closeness centrality.
Note: A list of characteristics of all 113 fisheries is available in the Appendix.
Fishery (gear-
area-species
code)
Federally
Managed
Average
Number of
Permits
Degree
Average Edge
Weight
Closeness
Centrality
Cluster
Statewide (CFEC area code = B)
Halibut – longline (B-6/61-B)
Y 2,261
10
6
34 0.96 1
Groundfish – longline (M-6/61-B)
Y
432
98
12
0.89
1
Sablefish – longline (C-6/61-B)
Y
625
85
18
0.80
1
Groundfish – jig (M-26-B)
N
404
80
9
0.78
2
Groundfish – pot (M-9/91-B)
Y
247
70
9
0.73
2
Salmon – troll (S-15-B)
N
961
67
16
0.71
1
Salmon – hand troll (S-5-B)
N
1,038
60
11
0.68
1
Halibut jig (B-26-B)
N
53
51
2
0.65
2
Groundfish – hand troll (M-5-B)
N
57
51
1
0.65
1
Groundfish trawl (M-7/71-B)
Y
295
46
3
0.62
2
Regional
Salmon – gill net Bristol Bay (S-3-T)
N
1,844
93
7
0.86
4
Salmon – seine – Kodiak (S-1-K)
N
365
77
8
0.76
2
Salmon – seine – Southeast (S-1-A)
N
358
68
9
0.72
1
Sea Cucumber Dive Southeast (Q-11-A)
N
320
63
7
0.69
5
Shrimp – pot – Southeast (P-9/91-A)
N
273
63
9
0.69
1
Tanner Crab – pot – other (TB-9/91-AK)
N
175
62
8
0.69
2
Salmon – seine – AK Peninsula (S-1-M)
N
116
61
4
0.69
2
Roe herring gill net Norton Sound (G-
34-Z)
N 260 60 3 0.68 4
Salmon – seine Prince William Sound (S-
1-E)
N 262 60 6 0.68 3
Dungeness Crab – pot – Southeast (D-9/91-
A)
N 279 59 8 0.67 1
Resources for the Future Addicott et al. 2017
Table 3: Edge Properties. There are 1,994 undirected edges in the network, which represent 997
pairs of fisheries that shared at least one permit holder from 2006-2015. Aggregate network
statistics appear in the first row. In the remaining rows, the edges are grouped by area and
jurisdiction. There are two types of areas and two types of jurisdictions, so three combinations
each of jurisdiction and area pairs (order of the pairing does not matter). The percentage of total
edges falling into each area and jurisdiction category is given in parentheses.
Number of
Edges
Potential
Edges
N*(N-1)/2
Average
Edge
Weight
Entire Network
1,994
6,328
5.82
Fishing Region
Statewide-Statewide
153
(8%)
231
(4%)
17.67
Regional-Regional
1,068
(54%)
4,095
(65%)
3.70
Statewide-Regional
773
(39%)
2,002
(32%)
6.40
Management Entity
Federal-Federal
44
(2%)
45
(1%)
31.16
State-State
1,497
(75%)
5,151
(81%)
3.72
Federal-State
453
(23%)
1,177
(19%)
10.30
*Totals may not add to 100% due to rounding.
Resources for the Future Addicott et al. 2017
Table 4: Cluster Summary Statistics
Cluster Nodes
Average
Number
of
Permits
per
Fishery
Density Average
Degree
Average
Edge
Weight
Average
Closeness
Centrality
Most Central Cluster
Fisheries
1 37 228 .63 22.60 12.50 0.75
Halibut, sablefish, and
groundfish statewide
longline
2 33 151 .58 18.67 5.25 0.72
Federally managed
groundfish pot;
statewide jig
groundfish; statewide
octopus/squid pot
3 15 59 .61 8.53 7.48 0.72
Southeast area fisheries.
Top 4 are Prince
William Sound: herring
and shrimp pot; salmon
seine and gillnet
4 21 282 .42 8.38 7.30 0.63
West and Northwest
area fisheries.
5
5
103
1
4
15.25
1
Dive fisheries.
6 1 6 - - - -
Scallop fishery,
W22BV
7
1
13
-
-
-
-
Hair crab, E91QV
Resources for the Future Addicott et al. 2017
Table 5: Marginal effects of fishery characteristics on cluster membership
Probability of Sharing a Cluster
a
(a)
(b)
Characteristic
Share
Characteristic
Do not Share
Characteristic
Difference
(a) - (b)
P-value
Species
0.232
0.254
-0.0219
0.162
(0.0141)
(0.00633)
(0.0157)
Gear
0.336
0.228
0.108
0.000
(0.0146)
(0.00636)
(0.0162)
Area
0.549
0.162
0.387
0.000
(0.0150)
(0.00602)
(0.0162)
aAverage predicted probability from Logit regression model with standard
errors in the parentheses.
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Individuals relying on natural resource extraction for their livelihood face high income variability driven by a mix of environmental, biological, management, and economic factors. Key to managing these industries is identifying how regulatory actions and individual behavior affect income variability, financial risk, and, by extension, the economic stability and the sustainable use of natural resources. In commercial fisheries, communities and vessels fishing a greater diversity of species have less revenue variability than those fishing fewer species. However, it is unclear whether these benefits extend to the actions of individual fishers and how year-to-year changes in diversification affect revenue and revenue variability. Here, we evaluate two axes by which fishers in Alaska can diversify fishing activities. We show that, despite increasing specialization over the last 30 years, fishing a set of permits with higher species diversity reduces individual revenue variability, and fishing an additional permit is associated with higher revenue and lower variability. However, increasing species diversity within the constraints of existing permits has a fishery-dependent effect on revenue and is usually (87% probability) associated with increased revenue uncertainty the following year. Our results demonstrate that the most effective option for individuals to decrease revenue variability is to participate in additional or more diverse fisheries. However, this option is expensive, often limited by regulations such as catch share programs, and consequently unavailable to many individuals. With increasing climatic variability, it will be particularly important that individuals relying on natural resources for their livelihood have effective strategies to reduce financial risk.
Article
Full-text available
Marine social–ecological systems are constantly changing, and fishers who make a living from working the seas are continually adapting in response to different sources of variability. One main way in which fishers can adapt to ecosystem change is to change the fisheries they participate in. This acts to connect fisheries, creating interlinked networks of alternative sources of income for fishers. Here, we synthesize fisheries data and construct fisheries connectivity networks for all major ports in the US California Current Large Marine Ecosystem. Fisheries connec-tivity networks are comprised of nodes, which are fisheries, connected by edges, whose weights are proportional to the number of participating vessels. Fisheries connectivity networks identify central fisheries in the US California Current Large Marine Ecosystem, specifically Dungeness crab and Spiny Lobster, and systematic topological differences, e.g. in network resilience and modularity. These network metrics directly relate to the social vulnerability of coastal fishing communities, especially their sensitivity and capacity to adapt to perturbation. Ultimately, improving knowledge of fisheries connectivity is vital if policy makers are to create governance institutions that allow fishermen to adapt to environmental, technological and management change while at the same time enhancing the social and economic value of fisheries. In doing so, new policies that account for fisheries connectivity, will lead to improved sustainable fisheries management, and enhanced socioeconomic resilience of coastal communities.
Article
Full-text available
Abrupt shifts in natural resources and their markets are a ubiquitous challenge to human communities. Building resilient social-ecological systems requires approaches that are robust to uncertainty and to regime shifts. Harvesting diverse portfolios of natural resources and adapting portfolios in response to change could stabilize economies reliant on natural resources and their markets, both of which are prone to unpredictable shifts. Here we use fisheries catch and revenue data from Alaskan fishing communities over 34 years to test whether diversification and turnover in the composition of fishing opportunities increased economic stability during major ocean and market regime shifts in 1989. More than 85% of communities show reduced fishing revenues following these regime shifts. However, communities with the highest portfolio diversity and those that could opportunistically shift the composition of resources they harvest, experienced negligible or even positive changes in revenue. Maintaining diversity in economic opportunities and enabling turnover facilitates sustainability of communities reliant on renewable resources facing uncertain futures.
Article
Full-text available
Social networks can profoundly affect human behavior, which is the primary force driving environmental change. However, empirical evidence linking microlevel social interactions to large-scale environmental outcomes has remained scarce. Here, we leverage comprehensive data on information-sharing networks among large-scale commercial tuna fishers to examine how social networks relate to shark bycatch, a global environmental issue. We demonstrate that the tendency for fishers to primarily share information within their ethnic group creates segregated networks that are strongly correlated with shark bycatch. However, some fishers share information across ethnic lines, and examinations of their bycatch rates show that network contacts are more strongly related to fishing behaviors than ethnicity. Our findings indicate that social networks are tied to actions that can directly impact marine ecosystems, and that biases toward within-group ties may impede the diffusion of sustainable behaviors. Importantly, our analysis suggests that enhanced communication channels across segregated fisher groups could have prevented the incidental catch of over 46,000 sharks between 2008 and 2012 in a single commercial fishery.
Article
Full-text available
In 2011 the National Marine Fisheries Service began a systematic collection of performance indicators for U.S. fisheries managed under catch shares. Catch shares are a fishery management tool that dedicate a secure share of quota allowing individual fishermen, fishing cooperatives, fishing communities, or other entities to harvest a fixed amount of fish. Catch share design varies widely across different programs and regions. Many programs share similar biological, social, and economic management objectives even though these design features are tailored to accommodate particular fishery characteristics. This paper evaluates fisheries using standardized indicators to measure the basic economic performance, regardless of catch share program design. Data collected were used to evaluate the economic and distribution effects of U.S. catch share programs. Catch share fishery performance is compared to a baseline period prior to implementation of the catch share program. Overall, the majority of objectives to improve the economic performance of catch share fisheries were achieved. Catch share programs have been effective in reducing fishing capacity. However, catch share programs have had distributional consequences as there are indications that consolidation is occurring in a number of programs. For example, there have been considerable reductions in the number of active vessels and entities holding quota share in the Alaska Halibut and Sablefish and the Mid-Atlantic Surfclam and Ocean Quahog catch share programs. However, it is important to note that the accumulation of ownership share may be less of a concern than consolidation in the use of quota. Thus, to the extent that consolidation is considered a management problem, it may be more effective to consider caps on the use of quota than by imposing more restrictive ownership caps.
Article
Full-text available
The food production system is increasingly global and seafood is among the most highly traded commodities. Global trade can improve food security by providing access to a greater variety of foods, increasing wealth, buffering against local supply shocks, and benefit the environment by increasing overall use efficiency for some resources. However, global trade can also expose countries to external supply shocks and degrade the environment by increasing resource demand and loosening feedbacks between consumers and the impacts of food production. As a result, changes in global food trade can have important implications for both food security and the environmental impacts of production. Measurements of globalization and the environmental impacts of food production require data on both total trade and the origin and destination of traded goods (the network structure). While the global trade network of agricultural and livestock products has previously been studied, seafood products have been excluded. This study describes the structure and evolution of the global seafood trade network, including metrics quantifying the globalization of seafood, shifts in bilateral trade flows, changes in centrality and comparisons of seafood to agricultural and industrial trade networks. From 1994 to 2012 the number of countries trading in the network remained relatively constant, while the number of trade partnerships increased by over 65%. Over this same period, the total quantity of seafood traded increased by 58% and the value increased 85% in real terms. These changes signify the increasing globalization of seafood products. Additionally, the trade patterns in the network indicate: increased influence of Thailand and China, strengthened intraregional trade, and increased exports from South America and Asia. In addition to characterizing these network changes, this study identifies data needs in order to connect seafood trade with environmental impacts and food security outcomes.
Article
Managing ecosystems is challenging because of the high number of stakeholders, the permeability of man-made political and jurisdictional demarcations in relation to the temporal and spatial extent of biophysical processes, and a limited understanding of complex ecosystem and societal dynamics. Given these conditions, collaborative governance is commonly put forward as the preferred means of addressing environmental problems. Under this paradigm, a deeper understanding of if, when, and how collaboration is effective, and when other means of addressing environmental problems are better suited, is needed. Interdisciplinary research on collaborative networks demonstrates that which actors get involved, with whom they collaborate, and in what ways they are tied to the structures of the ecosystems have profound implications on actors’ abilities to address different types of environmental problems.
Article
Managing ecosystems to provide ecosystem services in the face of global change is a pressing challenge for policy and science. Predicting how alternative management actions and changing future conditions will alter services is complicated by interactions among components in ecological and socioeconomic systems. Failure to understand those interactions can lead to detrimental outcomes from management decisions. Network theory that integrates ecological and socioeconomic systems may provide a path to meeting this challenge. While network theory offers promising approaches to examine ecosystem services, few studies have identified how to operationalize networks for managing and assessing diverse ecosystem services. We propose a framework for how to use networks to assess how drivers and management actions will directly and indirectly alter ecosystem services.
Article
Regional councils manage U.S. fisheries. Fishermen can participate in fisheries managed by multiple councils, and effort controls in one region could lead to effort leakage into another. Theoretical modeling demonstrates that positive, negative, and no leakage are possible. Using difference-in-differences, we test for leakage across regional boundaries for a catch share program in New England and find evidence that the New England groundfish sector program caused spillover into adjacent Mid-Atlantic fisheries. Aggregate Mid-Atlantic harvest volume increased among sector members after the policy change. We find leakage in individual fisheries with similar gear and high market substitutability with sector species. © 2016 by the Board of Regents of the University of Wisconsin System.