Content uploaded by Adam Ralph Zeilinger
Author content
All content in this area was uploaded by Adam Ralph Zeilinger on Sep 27, 2017
Content may be subject to copyright.
Museum specimen data reveal emergence of a plant disease may
be linked to increases in the insect vector population
ADAM R. ZEILINGER,
1,2,6
GIOVANNI RAPACCIUOLO,
1,3
DANIEL TUREK,
4
PETER T. OBOYSKI,
5
RODRIGO P. P. A LMEIDA,
2
AND GEORGE K. RODERICK
1,2
1
Berkeley Initiative for Global Change Biology, University of California Berkeley,
3101 Valley Life Sciences Building, Berkeley, California 94720 USA
2
Department of Environmental Science, Policy, and Management, University of California Berkeley,
130 Mulford Hall, Berkeley, California 94720 USA
3
Stony Brook University, 650 Life Sciences Building, Stony Brook, New York 11789 USA
4
Department of Mathematics and Statistics, Williams College, Williamstown, Massachusetts 01267 USA
5
Essig Museum of Entomology, University of California Berkeley, 1101 Valley Life Sciences Building, Berkeley, California 94720 USA
Abstract. The emergence rate of new plant diseases is increasing due to novel introduc-
tions, climate change, and changes in vector populations, posing risks to agricultural sustain-
ability. Assessing and managing future disease risks depends on understanding the causes of
contemporary and historical emergence events. Since the mid-1990s, potato growers in the
western United States, Mexico, and Central America have experienced severe yield loss from
Zebra Chip disease and have responded by increasing insecticide use to suppress populations
of the insect vector, the potato psyllid, Bactericera cockerelli (Hemiptera: Triozidae). Despite
the severe nature of Zebra Chip outbreaks, the causes of emergence remain unknown. We
tested the hypotheses that (1) B. cockerelli occupancy has increased over the last century in
California and (2) such increases are related to climate change, specifically warmer winters. We
compiled a data set of 87,000 museum specimen occurrence records across the order Hemi-
ptera collected between 1900 and 2014. We then analyzed changes in B. cockerelli distribution
using a hierarchical occupancy model using changes in background species lists to correct for
collecting effort. We found evidence that B. cockerelli occupancy has increased over the last
century. However, these changes appear to be unrelated to climate changes, at least at the scale
of our analysis. To the extent that species occupancy is related to abundance, our analysis pro-
vides the first quantitative support for the hypothesis that B. cockerelli population abundance
has increased, but further work is needed to link B. cockerelli population dynamics to Zebra
Chip epidemics. Finally, we demonstrate how this historical macro-ecological approach
provides a general framework for comparative risk assessment of future pest and insect vector
outbreaks.
Key words: Bayesian analysis; Candidatus Liberibacter solanacearum; list length analysis; NIMBLE;
occupancy model; opportunistic ecological data.
INTRODUCTION
The emergence rate of novel infectious diseases of
humans, wildlife, and plants has increased dramatically
in recent decades (Anderson et al. 2004, Jones et al.
2008). For plant diseases, the majority of recent out-
breaks have been caused by the introduction of patho-
gens into new areas, but climate change and agricultural
change have also been implicated in recent emergence
events (Anderson et al. 2004). For vector-borne plant
pathogens, these same processes could also cause disease
outbreaks indirectly through changes in vector popula-
tions: evolutionary, numerical, or both (Anderson et al.
2004, Canto et al. 2009, Fereres 2015). Improved under-
standing of the causes of disease outbreaks is critical for
managing current epidemics and for assessing risks of
future ones (Garrett et al. 2011).
Climate change has the potential to exacerbate agri-
cultural disease risk in the near future and has already
caused contemporary outbreaks (Anderson et al. 2004,
Garrett et al. 2011). Much of the work on agricultural
pest and disease risks from climate change has relied on
a combination of small-scale experiments and modeling
to extrapolate experimental findings to larger scales
(Juroszek and von Tiedemann 2013). However, the mul-
tiple dimensions of climate change and the multitude of
biotic interactions have led to concerns about the rele-
vance of small-scale experiments (Van Der Putten et al.
2010, Juroszek and von Tiedemann 2013). At the same
time, historical macro-ecological analyses using long-
term data sets can contribute critical insight into the
possible role that climate change plays in contemporary
outbreaks amid the complexity of climatic variation and
interacting species (Jeger and Pautasso 2008).
Manuscript received 7 November 2016; revised 2 March
2017; accepted 12 April 2017. Corresponding Editor: Matthew
P. Ay r e s .
6
E-mail: arz@berkeley.edu
1827
Ecological Applications, 27(6), 2017, pp. 1827–1837
©2017 by the Ecological Society of America
Beginning in the mid-1990s, potato growers in western
United States, Mexico, and Central America saw wide-
spread losses from the emergence of Zebra Chip disease,
caused by infections of the bacterial pathogen Candida-
tus Liberibacter solanacearum (Lso; =C. L. psyllau-
rous) (Munyaneza et al. 2007, Hansen et al. 2008).
Zebra Chip disease symptoms include plant stunting,
leaf scorching, and tuber necrosis, progressing eventually
to plant decline and death (Sengoda et al. 2009). Lso
also causes severe losses in tomato and pepper produc-
tion (Liu and Trumble 2006, Brown et al. 2010). Since
its emergence, the management of Lso has primarily
involved suppression of its vector, Bactericera cockerelli
(Sulc) (Hemiptera: Triozidae) or potato psyllid, through
insecticide use (Butler and Trumble 2012, Guenthner
et al. 2012).
Despite increasing harm to agricultural production
and the environment from ongoing Lso outbreaks and
increasing insecticide applications (Rosson et al. 2006,
Guenthner et al. 2012), the causes of Lso emergence
remain largely unknown. While B. cockerelli is native to
the western United States and Mexico, Zebra Chip
disease was first described in 1994 and Lso in 2008
(Munyaneza et al. 2007, Hansen et al. 2008). And while
other hypotheses warrant further investigation (see
Discussion), one dominant hypothesis has emerged: that
greater overwintering survival of northern B. cockerelli
populations is causing Zebra Chip outbreaks.
Historically, the range of B. cockerelli was thought to
be largely confined by its relatively narrow thermal toler-
ance and the availability of host plants. Pletsch (1947)
and Wallis (1955) hypothesized that populations were
limited in northern latitudes by cold winter temperatures,
a lack of host plants during winter, or both; these regions
would be subsequently colonized by populations migrat-
ing from the south in springtime. Beginning in 2011, pop-
ulations of B. cockerelli were found overwintering in
Oregon, Washington, and Idaho (Horton et al. 2015),
well beyond the species’hypothesized historic northern
limits (Wallis 1955), although a lack of historical moni-
toring leaves it unclear whether this represents a recent
range expansion. Greater overwintering survival of north-
ern populations would likely cause much greater densities
in potato fields earlier in potato plant development, when
plants are most susceptible to Lso infection (Gao et al.
2009). More generally, climate change is expected to
increase overwintering survival of insect herbivores across
temperate regions (Bale et al. 2002, Klapwijk et al. 2012).
Ecological theory predicts that increases in vector den-
sity can drive outbreaks of endemic pathogens. While the
relationship between vector density and host infection
prevalence is complex and rarely linear, it is generally a
positive relationship (Madden et al. 2000). Vector-borne
pathogens that are persistently and vertically transmitted,
which includes Lso (Hansen et al. 2008, Sengoda et al.
2014), should exhibit distinct epidemic thresholds across
a range of vector population densities (Jeger et al. 1998),
with the pathogen maintaining low prevalence in the host
population until vector populations reach very high den-
sities (Madden et al. 2000). Early work has failed to find
a relationship between adult B. cockerelli density and Lso
infection prevalence, but has found positive relationships
between late-stage nymph density, adult infection preva-
lence, and Zebra Chip disease prevalence (Goolsby et al.
2012, Workneh et al. 2013). Furthermore, contemporary
Lso infection prevalence in B. cockerelli populations
appearstobeverylow,between0%and8%acrossthe
midwestern and northwestern United States (Goolsby
et al. 2012, Swisher et al. 2013). At such low infection
rates, very large vector populations may be required to
cause a disease epidemic.
While Zebra Chip disease was first described in the
1990s, there is some evidence that Lso has coexisted with
B. cockerelli for much longer than Zebra Chip disease
has been recognized (see Discussion). Historically, Zebra
Chip may have been confused with pathological effects
of B. cockerelli saliva, known as Psyllid Yellows disease,
first described in the 1920s (Richards 1928). Given that
visual foliar symptoms of Psyllid Yellows and Zebra
Chip disease are quite similar (Munyaneza et al. 2007,
Sengoda et al. 2009) and that Lso is unculturable, earlier
researchers may have observed a combination of patho-
logical effects from B. cockerelli saliva and Lso infec-
tion. As a result, it is possible that Lso is endemic to
North America, has only occasionally erupted in out-
breaks, such as the Psyllid Yellows outbreaks of the
1920s and 1930s and the contemporary Zebra Chip out-
breaks, but remained largely undetected otherwise.
In the present paper, we tested whether numerical
changes in B. cockerelli populations could explain the
emergence of Zebra Chip disease. Specifically, we
hypothesize that (1) B. cockerelli populations have
increased over the last century in California, USA and
(2) such increases are related to climate change, specifi-
cally higher winter temperatures. We tested these
hypotheses by fitting museum specimen occurrence data
and historical climate projections to a hierarchical occu-
pancy model. Interest in historical ecological data, such
as that from natural history collections, has increased in
recent years due to their value in assessing impacts of cli-
mate change on disease dynamics and other ecological
processes (Tingley and Beissinger 2009). For example,
previous studies have used museum specimens to chart
invasion of the Lyme’s disease pathogen, Borrelia
burgdorferi, in tick vector specimens (Persing et al.
1990), changes in pathogenic fungal populations related
to air pollution in wheat specimens (Bearchell et al.
2005), and historical prevalence of chytrid fungus in
amphibian populations (Ouellet et al. 2005). At the
same time, museum specimens were often opportunisti-
cally collected by a diverse set of professional and citizen
naturalists with different levels of expertise and effort
(Isaac and Pocock 2015). They also represent presence-
only occurrence data, with little or no information about
where a species was searched for but not found. As a
result, museum specimen occurrence data are rife with
1828 ADAM R. ZEILINGER ET AL. Ecological Applications
Vol. 27, No. 6
biases, making rigorous ecological inference difficult
(Newbold 2010).
Hierarchical statistical models and Bayesian analytical
methods have facilitated the analysis of opportunistically
collected ecological data by explicitly modeling the
observation process separate from the ecological process
(Isaac et al. 2014). For example, occupancy models com-
bine an ecological sub-model, relating the presence and
absence, or occupancy, of a focal species to a set of
covariates, with a detection sub-model that relates detec-
tion to a distinct set of covariates. Following previous
work, we generated non-detection data (i.e., absence
data) for B. cockerelli and corrected for variation in col-
lecting effort using list length analysis (Phillips et al.
2009, Szabo et al. 2010). List length analysis involves
compiling lists of related species collected in the same
time and place and using the length of these lists as a
proxy for collecting effort. Some lists will contain the
focal species while many others will not; these later lists
constitute non-detection data. Because list length acts as
a proxy for overall collecting effort (Roberts et al. 2007),
incorporating list length as a covariate into a detection
sub-model provides estimates of detection probability
for the focal species, i.e., longer lists without the focal
species are more likely than shorter lists to represent true
absences. Similar analyses have been used to detect
trends in bird populations (Link et al. 2006, Szabo et al.
2010, Barnes et al. 2015), butterfly and dragonfly popu-
lations (Van Strien et al. 2010, Breed et al. 2012), and
changes in owl prey populations (Van Strien et al. 2015).
Here we show that the analysis of museum specimen
occurrence data using Bayesian hierarchical models can
provide insights into the causes of recent outbreaks of
agricultural pests and disease vectors. Additionally, in
the Discussion we outline how such a historical macro-
ecological approach to the analysis of disease vector and
pest outbreaks could facilitate development of risk
assessment frameworks for future outbreaks.
MATERIALS AND METHODS
Compiling species occurrence records
We first collected, digitized, and georeferenced all
pinned museum specimens of B. cockerelli from six
major California natural history museums: Bohart
Museum of Entomology, University of California Davis;
California Academy of Science, San Francisco; Califor-
nia State Collection of Arthropods (CSCA), Plant Pest
Diagnostics Center, California Department of Agricul-
ture; Essig Museum of Entomology, UC Berkeley; Los
Angeles County Museum of Natural History; and UC
Riverside Entomology Research Museum. For each col-
lection, we identified B. cockerelli specimens that had
not yet been determined to the species level, housed in
the “Undetermined Psyllidae”or “Undetermined Triozi-
dae”sections of the collections. We identified undeter-
mined specimens using the morphological description for
Paratrioza (=Bactericera)cockerelli in Tuthill (1945). We
also verified identification of a random subset of speci-
mens previously identified as B. cockerelli; all were con-
firmed correctly identified. We digitized all B. cockerelli
specimens, and georeferenced them using the point-
radius method of (Wieczorek et al. 2004) and the
GEOLocate web application.
7
In addition to pinned spec-
imens, we also included slide-mounted nymph specimens
from the Bohart Museum and records of ethanol-
preserved specimens from the Essig Museum. All georefer-
enced specimens, except for ethanol-preserved specimens,
were uploaded to the Essig Museum online database.
8
Our data set of B. cockerelli museum specimens repre-
sented presence-only data. To generate non-detection
data, we used the target-group background approach of
Phillips et al. (2009). When using presence-only data,
species distribution modeling requires the selection of
non-detection data, also known as background or
quadrature points in the spatial statistical literature
(Renner et al. 2015). Often non-detection points are
selected randomly or uniformly over the geographical or
environmental space of interest. However, Phillips et al.
(2009) selected non-detection points by using the pres-
ence of a larger set of related taxa (i.e., target group),
effectively reducing bias in the species distribution esti-
mates compared to uniform or random non-detection
selection methods.
We chose as our target group all species in the order
Hemiptera, excluding families that were predominantly
aquatic or carnivorous. The excluded families were Antho-
coridae, Belostomatidae, Cimicidae, Corixidae, Gelasto-
coridae, Geocoridae, Gerridae, Hebridae, Hydrometridae,
Leptopodidae, Macroveliidae, Mesoveliidae, Nabidae,
Naucoridae, Nepidae, Notonectidae, Ochteridae, Phymati-
dae, Pieidae, Polyctenidae, Reduviidae, Saldidae, and
Veliidae. The remaining families of Hemiptera are pre-
dominately (though not exclusively) composed of terres-
trial piercing-sucking herbivores and are likely to be
collected by entomologists in similar manners.
To generate our target group data set, we compiled
digitized and georeferenced California records of our
selected hemipteran families from three sources: the
Global Biodiversity Information Facility (GBIF; which
includes all records from the Essig Museum online data-
base), the American Museum of Natural History
(AMNH), and the CSCA database. An initial explo-
ration of these databases returned only 25 records of
B. cockerelli, far too few for analysis and thus requiring
the digitization and georeferencing of additional speci-
mens described above. In the CSCA database, we
included only records that were opportunistically col-
lected, excluding those that reported trap monitoring for
specific pest species. The data from the CSCA were
biased toward agricultural areas; however, this partially
balanced biases against agricultural areas in the GBIF
7
http://www.museum.tulane.edu/geolocate/default.html
8
https://essigdb.berkeley.edu/
September 2017 PEST OUTBREAKS AND MUSEUM SPECIMEN DATA 1829
and AMNH data sets. Occurrence records were included
in our data set if they were collected within California
between 1900 and 2014, included the full species name,
and were georeferenced.
From our compiled records, we generated lists of spe-
cies collected across the same time and area of our
B. cockerelli records. To do so, we divided California
into 15 915 km grid cells and combined all species
records collected within the same spatial cell, month,
and year. This spatial resolution was similar to that pre-
viously used by Hill et al. (1999) to model changes in
distribution of butterfly distribution under climate
change. We included only lists with three or more spe-
cies, following methods of Van Strien et al. (2013), and
lists in which at least one collector had also collected a
B. cockerelli specimen in our data set. By filtering lists to
a set of common collectors, we reduced the influence of
collectors focused on only a single family or other taxon,
a common concern in the analysis of opportunistically
collected data (Isaac and Pocock 2015).
Climate data
Our aim was to model the temporal trends in
B. cockerelli occupancy and to test if these trends could
be explained by local-scale climate change. As such, we
compiled estimates of historical climate from the Califor-
nia Basin Characterization Model (BCM), which pro-
vides estimates of a range of temperature- and
precipitation-derived climatic indices at 270 9270 m
grid resolution for much of California on a monthly basis
from 1895 to 2014 (Flint et al. 2013, Flint and Flint
2014). For each species occurrence record in a list, we
extracted estimates of actual evapotranspiration (AET)
in the month of collection, the water-year annual mini-
mum temperature (T
min
,°C), and the water-year annual
maximum temperature (T
max
,°C). For annual minimum
and maximum temperatures, we extracted the minimum
temperature occurring in the most recent winter (Decem-
ber–February) and most recent summer (June–Septem-
ber), respectively. Because our species lists were compiled
at a larger spatial cell size than the BCM cells, we
extracted the climate values for each occurrence record
within a list and then averaged these estimates to obtain
a list-level estimate of climate. The standard deviations
of these averages across lists were small for each of the
three climate variables (AET mean SD =2.08; T
min
mean
SD =0.17; T
max
mean SD =0.26) relative to their stan-
dard deviations across all specimen records in the data
set (AET =35.71, T
min
=4.01, T
max
=5.86), indicating
that our averaging of climate variables at the list level did
not neglect significant within-list climatic variation.
AET is an estimate of the water available for plant
growth; it is a function of precipitation and temperature,
and is more biologically relevant than raw precipitation
estimates (Rapacciuolo et al. 2014). AET has been used
previously to successfully model changes in herbivorous
insect distributions from climate change (Hill et al.
1999). Additionally, historical descriptions of the distri-
bution of B. cockerelli indicate that the species primarily
occurs in arid and semiarid climates of the southwestern
and Rocky Mountain regions of the United States (Wallis
1955), which would be captured by estimates of AET. We
also considered incorporating estimates of climate water
deficit (CWD); however, AET and CWD were highly neg-
atively correlated within our data set and we thus opted
for keeping the former only (Appendix S1: Fig. S1).
We included annual T
min
because the prevailing
hypothesis explaining the emergence of Zebra Chip Dis-
ease is that B. cockerelli populations at the northern lim-
its of the species range have increased due to greater
overwintering survival (Wallis 1955, Horton et al. 2015).
If true, then B. cockerelli occupancy should be positively
associated with annual T
min
. Finally, we included annual
T
max
because of historical descriptions of B. cockerelli
phenology suggesting that southern populations are extir-
pated due to high summer temperatures (Romney 1939).
Statistical analyses
We modeled B. cockerelli occupancy using a hierarchi-
cal occupancy model, also known as a binomial–bino-
mial mixture model, as described in Royle and K
ery
(2007). This model explicitly models the observation or
detection process separately from the ecological or occu-
pancy processes that give rise to species occupancy pat-
terns and corrects for zero-inflation in our data set
(Martin et al. 2005). Importantly, however, our data did
not include repeated visits to a site. Rather, by using
non-overlapping sets of covariates for the two sub-
models, the occupancy and detection latent states were
identifiable (S
olymos et al. 2012). For the occupancy sub-
model, we included the year collected, AET, T
min
,T
max
,
and a quadratic form for month collected. For the detec-
tion sub-model, we included list length and the interac-
tion between list length and year collected. We included
the interaction between list length and year to correct for
changes in the relationship between list length and occu-
pancy over time due to, e.g., changing collecting and pre-
serving technologies and collector expertise. List length
was natural-log-transformed (Szabo et al. 2010) and all
continuous covariates were standardized around the
mean and divided by the standard deviation. To incorpo-
rate spatial autocorrelation, we included the 15 915 km
spatial grid cells used to construct species lists as a ran-
dom effect in the occupancy sub-model. In other words,
lists that originated from the same location (i.e., spatial
grid cell) were assigned the same intercept. According to
the simulations of Isaac et al. (2014), occupancy models
with list length and a spatial random effect can be slightly
conservative, but are overall more robust to bias in
opportunistic occurrence data than other models.
We fit the occupancy model using the Markov chain
Monte Carlo (MCMC) engine provided with the NIM-
BLE package (version 0.6-3) for R 3.3.2 (NIMBLE
Development Team 2015, De Valpine et al. 2016, R Core
1830 ADAM R. ZEILINGER ET AL. Ecological Applications
Vol. 27, No. 6
Team 2016). MCMC speed and convergence were expe-
dited by making use of the flexibility of NIMBLE algo-
rithms. Latent states representing true (unknown) site
occupancy were analytically removed from the model
formulation using a custom-specified distribution, as
described in Turek et al. (2016). Block sampling was
used to jointly sample the coefficients of each linear pre-
dictor term within each sub-model, since these will gen-
erally exhibit strong posterior correlation (Roberts and
Sahu 1997). In addition, the standard deviation of site
random effects was sampled on a logarithmic scale using
the generalized Gibbs sampling framework described in
Liu and Sabatti (2000). We used uninformative priors
and three MCMC chains each with 150,000 iterations
and a burn-in period of 50,000. Convergence was veri-
fied by calculating Gelman-Rubin diagnostic and effec-
tive sample size (Gelman and Rubin 1992, Brooks and
Gelman 1998). R code for fitting models is available in
DataS1.zip; additional code for compiling and filtering
data and analyzing model output can be found at
github.com (https://doi.org/10.5281/zenodo.555994).
RESULTS
Our final data set of specimen records included 87,035
records of 2,840 species from 73 families; it also included
613 records of B. cockerelli. Specimens of all selected
Hemiptera were collected throughout California, with
the greatest number collected in the 1930s, 1960s, and
1970s (Fig. 1). Specimens of B. cockerelli were concen-
trated in southern California and arid interior regions of
the state; the majority of specimens were collected from
1950 to 1970 with the first and last B. cockerelli speci-
men collected in 1908 and 2011, respectively (Fig. 1).
From this specimen data set, we assembled 900 species
lists, which ranged in length from 3 to 40 species, 36 of
which contained B. cockerelli (Appendix S1: Fig. S2).
The MCMC chains successfully converged. Estimates
of the Gelman-Rubin diagnostic for all parameters were
≤1.1 and effective sample sizes of all parameters were
>1,000 (Table 1).
For the detection sub-model, list length was margin-
ally significant and positively related to probability of
detection, as was the list length 9year interaction
(Table 1, Fig. 2). The marginally significant interaction
term was likely due to a small but significant decline in
list length over time (Appendix S1: Fig. S3).
For the occupancy sub-model, the year main effect
was significant and positive (Table 1). However, none of
the climate variables were significant. The model pre-
dicts an increase in B. cockerelli probability of occu-
pancy in the late autumn/early winter in the 1950s and
increases in B. cockerelli throughout all seasons begin-
ning in the 1990s, with particularly high occupancy rates
in summers (Fig. 2).
We generated maps of the estimated occupancy proba-
bility over three different time periods: 1920–1945,
1950–1975, and 1990–2015 (Fig. 3), covering the
historical Psyllid Yellows outbreak, a non-outbreak per-
iod, and the contemporary outbreak period, respectively.
Although recent data are sparse, the model predicts an
overall increase in occupancy probability in southern
California in more recent years, with greater variation in
occupancy in early time periods.
DISCUSSION
Since emergence in the early 1990s, Zebra Chip disease
has devastated potato production in epidemic areas of
the United States, Mexico, and Central America (Secor
and Rivera-Varas 2004, Rosson et al. 2006, Horton et al.
FIG. 1. (a) Spatial and (b) temporal distribution of all
museum records (gray points and bars) and of Bactericera cock-
erelli museum records (black points and bars).
September 2017 PEST OUTBREAKS AND MUSEUM SPECIMEN DATA 1831
2015). For example, whereas Texas potato growers tradi-
tionally used few or no insecticides, growers currently
average over seven applications annually as a direct result
of Zebra Chip emergence (Goolsby et al. 2012, Guenth-
ner et al. 2012). Yet the causes of these outbreaks remain
unknown. Using museum specimen occurrence data and
a Bayesian occupancy model, we showed that occupancy
probability of the vector, B. cockerelli, has increased in
California over the last century.
Our analysis showed increasing occupancy probability
over the last century through a significant and positive
year main effect. However, sparse data weaken further
inference. Despite our initial large data set (>87,000 hemi-
pteran records and >600 B. cockerelli records), our final
data set consisted of 900 total species lists with only 36
containing B. cockerelli. The sharp reduction in data when
constructing the lists was due to a large number of single
species lists and to a large number of duplicate records,
which in each case reduces to a single detection event.
The reduced spatial coverage and total number of lists
in more recent decades, particularly in northern Califor-
nia, prevented us from testing for a potential northward
range expansion of B. cockerelli (Fig. 3). However, we
do not expect this decline in the number of records to
have influenced our main result of an overall increase in
occupancy probability for B. cockerelli over time.
Despite there being fewer lists in recent decades, the
proportion of those lists containing B. cockerelli, the
species’reporting rate (Szabo et al. 2010), increased
markedly (Appendix S1: Fig. S4). This trend persists
after correcting for overall collecting effort and changes
in collecting effort over time, which we accounted for by
including a list length main effect and a list-length 9
year interaction effect in the detection sub-model.
In epidemiology of vector-borne pathogens, patterns of
vector occupancy are less relevant than patterns of vector
population dynamics for determining pathogen spread.
Nonetheless, occupancy probability over large spatial
scales is positively related to abundance (Gaston et al.
2000, Oliver et al. 2012, Guti
errez et al. 2013) except
when species exhibit very low colonization ability (Freck-
leton et al. 2005). Evidence suggests that B. cockerelli
exhibits high colonization ability (Nelson et al. 2014).
Thus, a positive occupancy–abundance relationship
should hold and the increases in occupancy probability
that we detected should generally be associated with
increases in vector abundance. In future analyses, it may
be possible to treat duplicate species records as count
data and directly model abundance within a list length
analysis framework. However, theory relating species list
lengths to count data has not been developed and would
likely require additional statistical assumptions.
We had hypothesized that occupancy should be posi-
tively related to T
min
based on the previous literature
(Wallis 1955). While we found no evidence for an associ-
ation between T
min
and B. cockerelli occupancy, climate
change could influence B. cockerelli populations and
Lso emergence through processes beyond our present
analysis. Changes in degree-days are often more relevant
for insect development rates than annual minimum or
maximum temperatures (Bale et al. 2002). Recently,
Tran et al. (2012) showed that nymphs of B. cockerelli
ceased development and ~90% of individuals died at a
constant 8°C under laboratory conditions. In our data
set, T
min
reached below 8°C in 98% of all lists and 97%
of lists containing B. cockerelli. This apparent discrep-
ancy may be because the duration of cold temperatures
or the temperatures experienced over the entire develop-
mental period, as captured in degree-days, are more
important than the minimum temperature experienced
at a single point in time. We hope to incorporate degree-
days into future analyses.
Climate change could also influence B. cockerelli
populations indirectly. We hypothesized that any rela-
tionship between AET and B. cockerelli occupancy
would be mediated through changes in host plant popu-
lations. However, we found no evidence of a relationship
between AET and B. cockerelli occupancy. Explicitly
TABLE 1. Estimates of the Gelman-Rubin diagnostic, ^
r(with 97.5% credible intervals, CI), effective sample size (ESS), and
posterior mean (with 95% CI) for linear coefficients for Bactericera cockerelli occupancy model.
Model term ^
r(97.5% CI) ESS Mean [2.5% CI, 97.5% CI]
Year 1 (1) 2,268.39 50.4 [16.1, 95.4] ‡
Month 1 (1) 2,976.61 1.88 [48.7, 53.9]
Month
2
1 (1) 3,536.05 9.12 [40.9, 57.9]
AET 1 (1) 3,068.94 22.7 [73, 31.3]
T
min
1 (1) 2,470.83 13.6 [62.7, 35.2]
T
max
1 (1) 2,579.5 11.6 [62.9, 41.5]
Detection sub-model intercept 1 (1) 7,144.72 2.41 [2.86, 1.97] ‡
List length 1 (1) 24,529.71 0.283 [0.0478, 0.595] †
List length 9year 1 (1) 24,186.37 0.332 [0.0172, 0.701] †
l
a
1 (1) 5,088.08 11 [65.6, 48.1]
r
a
1 (1.01) 1,219.42 271 [57.4, 868]
Notes: AET, actual evapotranspiration; T
min
, water-year annual minimum temperature; T
max
, water-year annual maximum
temperature; l
a
, mean of the spatial cell random effect; r
a
, variance of the spatial cell random effect.
†Significant at 90% CI; ‡significant at 95% CI.
1832 ADAM R. ZEILINGER ET AL. Ecological Applications
Vol. 27, No. 6
modeling changes in host plant occupancy in relation to
climatic variation and then relating host plant occu-
pancy to B. cockerelli occupancy, perhaps through a
joint distribution model, would be a more robust test.
Finally, beyond any effects on B. cockerelli, climate
change could contribute to Lso outbreaks by directly
influencing Lso populations or the interactions between
host plant, vector, and pathogen (Garrett et al. 2006).
Separate from climate change, B. cockerelli popula-
tion abundances may have increased through other
mechanisms. Horton et al. (2015) have hypothesized that
the outbreaks in the northwestern United States are
caused by invasion of a nonnative host plant, Solanum
dulcamara (Solanaceae), and concomitant adaptation by
B. cockerelli populations. More broadly, the role of wild
host plants in B. cockerelli population dynamics and
distribution remains an important area for future
research. Additionally, population or genetic differences
in B. cockerelli reproduction, development, and insecti-
cide resistance have been described by previous authors,
any of which could contribute to Zebra Chip emergence
(Liu and Trumble 2007, Mustafa et al. 2015a, b).
In a review of plant disease emergence events by
Anderson et al. (2004), pathogen introduction was the
FIG. 2. Occupancy model results. (a) Relationships between occupancy model-predicted probability of detection and list length,
(b) between model-predicted probability of B. cockerelli occupancy and year collected, and (c) the relationships of occupancy prob-
ability, year collected, and month collected. Trend lines on scatterplots are linear model fits to model-predicted probabilities. For
the contour plot (c), grayscale values indicate probability of occupancy.
September 2017 PEST OUTBREAKS AND MUSEUM SPECIMEN DATA 1833
most common cause of emergence. For Zebra Chip dis-
ease, introduction of Lso into North America remains a
possible but unlikely explanation. Lso has recently been
detected in Europe and is associated with the triozid spe-
cies Bactericera trigonica and Trioza apicalis (Munyaneza
et al. 2010, Teresani et al. 2014). However, the European
isolates appear to be genetically distinct from the North
American isolates and infect primarily non-solanaceous
host plants (Nelson et al. 2011, Lin and Civerolo 2014).
We also constructed a cladogram of available 16S rDNA
sequences that supports previous assertions that an intro-
duction of Lso from Europe to North America has not
occurred recently (Appendix S2). Future genetic analyses
of historical and contemporary samples would provide
critical insight into the origins of Lso. Importantly, none
of the above possible causes are mutually exclusive.
Rather, multiple factors may be contributing to the emer-
gence of Lso-associated diseases and may differ in rela-
tive importance in different epidemic areas.
Understanding the causes of plant disease outbreaks is
critical for integrated management of pathogens and vec-
tors (Garrett et al. 2011). Our analysis has been the first
to show an increase in B. cockerelli occupancy over
relatively large spatial and temporal scales. However, more
work is needed to link B. cockerelli population dynamics
and distribution to Lso spread, especially given the diffi-
culty of detecting a relationship at smaller scales (Goolsby
et al. 2012, Workneh et al. 2013). Just as importantly,
future work on the underlying causes of B. cockerelli
population expansion will aid in identifying conditions
that generate the greatest risk of future epidemics.
The potential that climate change could reduce
agricultural sustainability through pests and disease out-
breaks, and resulting increases in pesticide use, has
become an increasing concern for growers, scholars, and
other stakeholders. As with ecological risk assessment of
other stressors, a key challenge in the analysis of outbreak
risks from climate change remains to identify which pest
or pathogen species are most likely to become more prob-
lematic and through which risk pathways (USEPA 1998,
Garrett et al. 2011). Most work to date has approached
the problem mechanistically, focused on predicting out-
break risks by testing for changes in particular biotic
interactions or responses from species with different life
history traits under a subset of predicted climatic changes
(Juroszek and von Tiedemann 2013). However, the
FIG. 3. Maps of estimated B. cockerelli occupancy across California, USA for three selected time periods, (a) 1920–1945, (b)
1950–1975, and (c) 1990–2015. Solid circles indicate species lists containing B. cockerelli (i.e., detection events); open circles indicate
non-detection events. The size of the circle indicates the estimated probability of occupancy, with larger circles representing greater
probabilities. Occupancy probabilities were averaged over years for each site. Map coordinates are the same as in Fig. 1a.
TABLE 2. Estimated posterior means (with 95% credible intervals) of linear coefficients for occupancy model of Lygus hesperus,
Myzus persicae, and Rhopalosiphum padi occurrence using the same model and data set as for B. cockerelli occupancy.
Model term Lygus hesperus Myzus persicae Rhopalosiphum padi
Year 52.8 [81.7, 23.1]‡7.91 [66.9, 54.3] 13.2 [54.9, 37.2]
Month 8.91 [4.72, 29.5] 24.1 [77, 32.1] 20.8 [67, 31.3]
Month
2
11.2 [29.2, 8.53] 7.34 [49.8, 38.8] 38.1 [12.9, 82.9]
AET 15.7 [0.0975, 36.8]†28.3 [28.2, 81.3] 21.4 [20.1, 72.2]
T
min
21.7 [53.4, 1.39]†1.6 [57.3, 58.5] 8.37 [57.8, 42.1]
T
max
13.9 [8.16, 33.9] 1.94 [54.1, 49.2] 3.02 [56.4, 57]
Detection intercept 1.44 [1.63, 1.24]‡5.79 [1.79, 22.2] 2.13 [2.59, 1.71]‡
List length 0.29 [0.124, 0.456]‡24.2 [0.611, 64.7]‡0.299 [0.0439, 0.641]†
List length 9year 0.0424 [0.188, 0.102] 19.7 [55.4, 0.753] 0.0141 [0.32, 0.35]
l
a
65.2 [33.1, 99.5] 16.4 [76.1, 49.2] 17.9 [42.6, 74.3]
r
a
63.5 [26, 121] 352 [79.7, 920] 362 [38, 944]
Note: MCMC convergence diagnostics showed similarly good convergence as for the B. cockerelli occupancy model.
†Significant at 90% CI; ‡significant at 95% CI.
1834 ADAM R. ZEILINGER ET AL. Ecological Applications
Vol. 27, No. 6
complexity of biological responses to climate change
limits the realism of small-scale mechanistic studies.
Our analysis points to an additional, complementary
approach. Modeling changes in occupancy for a range
of pest and vector species within our data set may pro-
vide valuable comparative insights into which species
pose the greatest risk. As an example, we applied the
same occupancy model to detection and non-detection
data for three pest species that were well represented in
the data set: Lygus hesperus (Hemiptera: Miridae), a pest
of strawberries and cotton; Myzus persicae (Hemiptera:
Aphididae), a vector of potato leafroll virus; and
Rhopalosiphum padi (Hemiptera: Aphididae), a vector of
barley yellow dwarf virus. Our analysis indicates that
L. hesperus occupancy probability has significantly
decreased in California over the last century, and is mar-
ginally related to changes in AET and T
min
. In contrast,
occupancy probabilities of M. persicae and R. padi have
shown no clear pattern of change (Table 2). Our results
for R. padi are in line with those of the analysis of Davis
et al. (2014) for populations in the northwestern United
States; in both cases, R. padi distribution showed little
temporal trend and no relationship to climatic variation.
Although our analysis is spatially and temporally
coarse and ignores the roles of biotic interactions and
agricultural change, comparisons among species, using a
common data set and model, could facilitate more fine-
grain investigations into the specific mechanisms that
underlie large-scale differences among pest species and
their relationships to climatic variation. Linking fine-
scale studies of biotic interactions and life history traits
to macro-ecological analyses will be necessary to develop
rigorous assessment frameworks for the future risks of
pest and pathogen outbreaks caused by climate change.
ACKNOWLEDGMENTS
We thank those who assisted us in sorting through and bor-
rowing specimens from entomological collections: Norm Penny,
California Academy of Sciences; Doug Yanega, UC Riverside
Entomology Research Museum; Lynn Kimsey, UC Davis
Bohart Museum of Entomology; Weiping Xie, LA County Nat-
ural History Museum; and Jackie Kishmirian, California State
Collection of Arthropods. Stephen Gaimari and Katja Selt-
mann helped with access to and filtering of the Plant Pest Diag-
nostic Center data and Christine Johnson helped with data
access from the American Museum of Natural History. Naia
Morueta-Holme and David Ackerly provided BCM climate
data. Emily Yao, Arsene Gaude, K Kai Din, Gordon Nishida,
and Joyce Gross assisted with digitizing and georeferencing
museum specimens. Sean Prager, Jay Rosenheim, Matt Daugh-
erty, members of the Almeida Lab and the Gillespie/Roderick
EvoLab, and two anonymous reviewers provided helpful com-
ments on earlier drafts. This work was supported by the Berke-
ley Initiative for Global Change Biology and the Gordon and
Betty Moore Foundation.
LITERATURE CITED
Anderson, P. K., A. A. Cunningham, N. G. Patel, F. J. Morales,
P. R. Epstein, and P. Daszak. 2004. Emerging infectious
diseases of plants: pathogen pollution, climate change and
agrotechnology drivers. Trends in Ecology & Evolution 19:
535–544.
Bale, J. S., et al. 2002. Herbivory in global climate change
research: direct effects of rising temperature on insect herbi-
vores. Global Change Biology 8:1–16.
Barnes, M., J. K. Szabo, W. K. Morris, and H. Possingham.
2015. Evaluating protected area effectiveness using bird lists
in the Australian Wet Tropics. Diversity and Distributions
21:368–378.
Bearchell, S. J., B. A. Fraaije, M. W. Shaw, and B. D. Fitt. 2005.
Wheat archive links long-term fungal pathogen population
dynamics to air pollution. Proceedings of the National Acad-
emy of Sciences USA 102:5438–5442.
Breed, G. A., S. Stichter, and E. E. Crone. 2012. Climate-driven
changes in northeastern US butterfly communities. Nature
Climate Change 3:142–145.
Brooks, S. P., and A. Gelman. 1998. General methods for
monitoring convergence of iterative simulations. Journal of
Computational and Graphical Statistics 7:434–455.
Brown,J.K.,M.Rehman,D.Rogan,R.R.Martin,andA.M.
Idris. 2010. First report of “Candidatus Liberibacter psyllau-
rous”(synonym “Ca. L. solanacearum”)associatedwith
“Tomato Vein-Greening”and “Tomat o P syl l i d Yel l ows ”diseases
in commercial greenhouses in Arizona. Plant Disease 94:376.
Butler, C. D., and J. T. Trumble. 2012. The potato psyllid, Bac-
tericera cockerelli (Sulc) (Hemiptera: Triozidae): life history,
relationship to plant diseases, and management strategies.
Terrestrial Arthropod Reviews 5:87–111.
Canto, T., M. A. Aranda, and A. Fereres. 2009. Climate change
effects on physiology and population processes of hosts and
vectors that influence the spread of hemipteran-borne plant
viruses. Global Change Biology 15:1884–1894.
Davis, T. S., J. T. Abatzoglou, N. A. Bosque-P
eRez, S. E.
Halbert, K. Pike, and S. D. Eigenbrode. 2014. Differing con-
tributions of density dependence and climate to the popula-
tion dynamics of three eruptive herbivores: cereal aphid
population dynamics. Ecological Entomology 39:566–577.
de Valpine, P., D. Turek, C. J. Paciorek, C. Anderson-Bergman,
D. T. Lang, and R. Bodik. 2016. Programming with models:
writing statistical algorithms for general model structures
with NIMBLE. Journal of Computational and Graphical
Statistics 1–28.
Fereres, A. 2015. Insect vectors as drivers of plant virus emer-
gence. Current Opinion in Virology 10:42–46.
Flint, L. E., and A. L. Flint. 2014. California Basin Characteri-
zation Model: a dataset of historical and future hydrologic
response to climate change. U.S. Geological Survey Data
Release. http://dx.doi.org/10.5066/F76T0JPB
Flint, L. E., A. L. Flint, J. H. Thorne, and R. Boynton. 2013.
Fine-scale hydrologic modeling for regional landscape
applications: the California Basin Characterization Model
development and performance. Ecological Processes 2:25.
Freckleton, R. P., J. A. Gill, D. Noble, and A. R. Watkinson.
2005. Large-scale population dynamics, abundance–occupancy
relationships and the scaling from local to regional population
size. Journal of Animal Ecology 74:353–364.
Gao, F., J. Jifon, X. Yang, and T.-X. Liu. 2009. Zebra chip dis-
ease incidence on potato is influenced by timing of potato
psyllid infestation, but not by the host plants on which they
were reared. Insect Science 16:399–408.
Garrett, K. A., S. P. Dendy, E. E. Frank, M. N. Rouse, and
S. E. Travers. 2006. Climate change effects on plant disease:
genomes to ecosystems. Annual Review of Phytopathology
44:489–509.
Garrett, K. A., et al. 2011. Complexity in climate-change
impacts: an analytical framework for effects mediated by
plant disease. Plant Pathology 60:15–30.
September 2017 PEST OUTBREAKS AND MUSEUM SPECIMEN DATA 1835
Gaston, K. J., T. M. Blackburn, J. J. D. Greenwood, R. D.
Gregory, R. M. Quinn, and J. H. Lawton. 2000. Abundance–
occupancy relationships. Journal of Applied Ecology 37:39–59.
Gelman, A., and D. B. Rubin. 1992. Inference from iterative simu-
lation using multiple sequences. Statistical Science 7:457–472.
Goolsby, J. A., J. J. Adamczyk Jr, J. M. Crosslin, N. N. Troxclair,
J. R. Anciso, G. G. Bester, J. D. Bradshaw, E. D. Bynum, L. A.
Carpio, and D. C. Henne. 2012. Seasonal population dynamics
of the potato psyllid (Hemiptera: Triozidae) and its associated
pathogen “Candidatus Liberibacter solanacearum”in potatoes
in the southern Great Plains of North America. Journal of
Economic Entomology 105:1268–1276.
Guenthner, J., J. Goolsby, and G. Greenway. 2012. Use and cost
of insecticides to control potato psyllids and zebra chip on
potatoes. Southwestern Entomologist 37:263–270.
Guti
errez,D.,J.Harcourt,S.B.D
ıez, J. G. Ill
an, and R. J. Wilson.
2013. Models of presence–absence estimate abundance as well
as (or even better than) models of abundance: the case of the
butterfly Parnassius apollo. Landscape Ecology 28:401–413.
Hansen, A. K., J. T. Trumble, R. Stouthamer, and T. D. Paine.
2008. A new Huanglongbing species, “Candidatus Liberibac-
ter psyllaurous”, found to infect tomato and potato, is vec-
tored by the psyllid Bactericera cockerelli (Sulc). Applied and
Environmental Microbiology 74:5862–5865.
Hill, J. K., C. D. Thomas, and B. Huntley. 1999. Climate and
habitat availability determine 20th century changes in a
butterfly’s range margin. Proceedings of the Royal Society B
266:1197–1206.
Horton, D. R., W. R. Cooper, J. E. Munyaneza, K. D. Swisher,
E. R. Echegaray, A. F. Murphy, S. I. Rondon, C. H. Wohleb,
T. D. Waters, and A. S. Jensen. 2015. A new problem and old
questions: potato psyllid in the Pacific Northwest. American
Entomologist 61:234–244.
Isaac, N. J. B., and M. J. O. Pocock. 2015. Bias and information
in biological records. Biological Journal of the Linnean
Society 115:522–531.
Isaac, N. J. B., A. J. van Strien, T. A. August, M. P. de Zeeuw,
and D. B. Roy. 2014. Statistics for citizen science: extracting
signals of change from noisy ecological data. Methods in
Ecology and Evolution 5:1052–1060.
Jeger, M. J., and M. Pautasso. 2008. Plant disease and global
change—the importance of long-term data sets. New Phytol-
ogist 177:8–11.
Jeger, M. J., F. van den Bosch, L. V. Madden, and J. Holt. 1998.
A model for analysing plant-virus transmission characteris-
tics and epidemic development. Mathematical Medicine and
Biology 15:1–18.
Jones, K. E., N. G. Patel, M. A. Levy, A. Storeygard, D. Balk,
J. L. Gittleman, and P. Daszak. 2008. Global trends in emerg-
ing infectious diseases. Nature 451:990–993.
Juroszek, P., and A. von Tiedemann. 2013. Plant pathogens,
insect pests and weeds in a changing global climate: a review
of approaches, challenges, research gaps, key studies and
concepts. Journal of Agricultural Science 151:163–188.
Klapwijk, M. J., M. P. Ayres, A. Battisti, and S. Larsson. 2012.
Assessing the impact of climate change on outbreak potential.
Pages 429–450 in P. Barbosa, D. K. Letourneau, and A. A.
Agrawal, editors. Insect outbreaks revisited. Wiley-Blackwell,
West S u s s ex, U K .
Lin, H., and E. L. Civerolo. 2014. Comparative genomics of the
liberibacteral plant pathogens. Pages 203–233 in D. C. Gross,
A. Lichens-Park, and C. Kole, editors. Genomics of plant-
associated bacteria. Springer, Berlin, Germany.
Link, W. A., J. R. Sauer, and D. K. Niven. 2006. A hierarchical
model for regional analysis of population change using
Christmas Bird Count data, with application to the American
Black Duck. Condor 108:13–24.
Liu, J., and C. Sabatti. 2000. Generalised Gibbs sampler and
multigrid Monte Carlo for Bayesian computation. Biome-
trika 87:353–369.
Liu, D., and J. Trumble. 2006. Ovipositional preferences, damage
thresholds, and detection of the tomato-potato psyllid Bac-
tericera cockerelli (Homoptera: Psyllidae) on selected tomato
accessions. Bulletin of Entomological Research 96:197–204.
Liu, D., and J. T. Trumble. 2007. Comparative fitness of invasive
and native populations of the potato psyllid (Bactericera cock-
erelli). Entomologia Experimentalis et Applicata 123:35–42.
Madden, L. V., M. J. Jeger, and F. van den Bosch. 2000. A theo-
retical assessment of the effects of vector-virus transmission
mechanism on plant virus disease epidemics. Phytopathology
90:576–594.
Martin, T. G., B. A. Wintle, J. R. Rhodes, P. M. Kuhnert, S. A.
Field, S. J. Low-Choy, A. J. Tyre, and H. P. Possingham.
2005. Zero tolerance ecology: improving ecological inference
by modelling the source of zero observations. Ecology Letters
8:1235–1246.
Munyaneza, J. E., J. M. Crosslin, and J. E. Upton. 2007. Associa-
tion of Bactericera cockerelli (Homoptera : Psyllidae) with “zebra
chip”, a new potato disease in southwestern United States and
Mexico. Journal of Economic Entomology 100:656–663.
Munyaneza, J. E., T. W. Fisher, V. G. Sengoda, S. F. Garczynski,
A. Nissinen, and A. Lemmetty. 2010. Association of “Candi-
datus Liberibacter solanacearum”with the psyllid, Trioza api-
calis (Hemiptera: Triozidae) in Europe. Journal of Economic
Entomology 103:1060–1070.
Mustafa, T., D. R. Horton, W. R. Cooper, K. D. Swisher, R. S.
Zack, and J. E. Munyaneza. 2015a. Interhaplotype fertility
and effects of host plant on reproductive traits of three
haplotypes of Bactericera cockerelli (Hemiptera: Triozidae).
Environmental Entomology 44:300–308.
Mustafa, T., D. R. Horton, K. D. Swisher, R. S. Zack, and J. E.
Munyaneza. 2015b. Effects of host plant on development and
body size of three haplotypes of Bactericera cockerelli (Hemi-
ptera: Triozidae). Environmental Entomology 44:593–600.
Nelson, W. R., T. W. Fisher, and J. E. Munyaneza. 2011. Haplo-
types of “Candidatus Liberibacter solanacearum”suggest long-
standing separation. European Journal of Plant Pathology
130:5–12.
Nelson, W. R., K. D. Swisher, J. M. Crosslin, and J. E.
Munyaneza. 2014. Seasonal dispersal of the potato psyllid,
Bactericera cockerelli, into potato crops. Southwestern Ento-
mologist 39:177–186.
Newbold, T. 2010. Applications and limitations of museum data
for conservation and ecology, with particular attention to species
distribution models. Progress in Physical Geography 34:3–22.
NIMBLE Development Team. 2015. NIMBLE: an r package
for programming with BUGS models, version 0.4. NIMBLE
Development Team https://CRAN.R-project.org/package=
nimble
Oliver, T. H., S. Gillings, M. Girardello, G. Rapacciuolo, T. M.
Brereton, G. M. Siriwardena, D. B. Roy, R. Pywell, and R. J.
Fuller. 2012. Population density but not stability can be pre-
dicted from species distribution models. Journal of Applied
Ecology 49:581–590.
Ouellet, M., I. Mikaelian, B. D. Pauli, J. Rodrigue, and D. M.
Green. 2005. Historical evidence of widespread chytrid infec-
tion in North American amphibian populations. Conserva-
tion Biology 19:1431–1440.
Persing, D. H., S. R. Telford, P. N. Rys, D. E. Dodge, T. J.
White, S. E. Malawista, and A. Spielman. 1990. Detection of
Borrelia burgdorferi DNA in museum specimens of Ixodes
dammini ticks. Science 249:1420–1423.
Phillips, S. J., M. Dud
ık, J. Elith, C. H. Graham, A. Lehmann,
J. Leathwick, and S. Ferrier. 2009. Sample selection bias and
1836 ADAM R. ZEILINGER ET AL. Ecological Applications
Vol. 27, No. 6
presence-only distribution models: implications for back-
ground and pseudo-absence data. Ecological Applications
19:181–197.
Pletsch, D. J. 1947. The potato psyllid, Paratrioza cockerelli
(Sulc), its biology and control. Bulletin of the Montana Agri-
cultural Experiment Station 446:1–95.
R Core Team. 2016. R: a language and environment for statisti-
cal computing. R Foundation for Statistical Computing,
Vienna, Austria.
Rapacciuolo, G., et al. 2014. Beyond a warming fingerprint:
individualistic biogeographic responses to heterogeneous
climate change in California. Global Change Biology 20:
2841–2855.
Renner, I. W., J. Elith, A. Baddeley, W. Fithian, T. Hastie, S. J.
Phillips, G. Popovic, and D. I. Warton. 2015. Point process
models for presence-only analysis. Methods in Ecology and
Evolution 6:366–379.
Richards, B. L. 1928. A new and destructive disease of the
potato in Utah and its relation to the potato psylla. Phy-
topathology 18:140–141.
Roberts, R. L., P. F. Donald, and R. E. Green. 2007. Using
simple species lists to monitor trends in animal populations:
new methods and a comparison with independent data.
Animal Conservation 10:332–339.
Roberts, G. O., and S. K. Sahu. 1997. Updating schemes, corre-
lation structure, blocking and parameterization for the Gibbs
sampler. Journal of the Royal Statistical Society: Series B
(Statistical Methodology) 59:291–317.
Romney, V. E. 1939. Breeding areas of the tomato psyllid, Par at ri oz a
cockerelli (
Sulc). Journal of Economic Entomology 32:150–151.
Rosson, P., M. Niemeyer, M. Palma, and L. Ribera. 2006.
Economic impacts of zebra chips on the Texas potato indus-
try. Center for North American Studies, Department of Agri-
cultural Economics, Texas A&M University, College Station,
Texas, USA.
Royle, J. A., and M. K
ery. 2007. A Bayesian state-space formu-
lation of dynamic occupancy models. Ecology 88:1813–1823.
Secor, G. A., and V. V. Rivera-Varas. 2004. Emerging diseases
of cultivated potato and their impact on Latin America.
Revista Latinoamericana de la Papa (Suplemento) 1:1–8.
Sengoda, V. G., W. R. Cooper, K. D. Swisher, D. C. Henne, and J.
E. Munyaneza. 2014. Latent period and transmission of “Can-
didatus Liberibacter solanacearum”by the potato psyllid Bac-
tericera cockerelli (Hemiptera: Triozidae). PLoS ONE 9:e93475.
Sengoda, V. G., J. E. Munyaneza, J. M. Crosslin, J. L.
Buchman, and H. R. Pappu. 2009. Phenotypic and etiological
differences between psyllid yellows and zebra chip diseases of
potato. American Journal of Potato Research 87:41–49.
S
olymos, P., S. Lele, and E. Bayne. 2012. Conditional likelihood
approach for analyzing single visit abundance survey data in
the presence of zero inflation and detection error. Environ-
metrics 23:197–205.
Swisher, K. D., V. G. Sengoda, J. Dixon, E. Echegaray, A. F.
Murphy, S. I. Rondon, J. E. Munyaneza, and J. M. Crosslin.
2013. Haplotypes of the potato psyllid, Bactericera cockerelli,
on the wild host plant, Solanum dulcamara, in the Pacific
Northwestern United States. American Journal of Potato
Research 90:570–577.
Szabo, J. K., P. A. Vesk, P. W. J. Baxter, and H. P. Possingham.
2010. Regional avian species declines estimated from
volunteer-collected long-term data using list length analysis.
Ecological Applications 20:2157–2169.
Teresani, G. R., et al. 2014. Association of “Candidatus Liberib-
acter solanacearum”with a vegetative disorder of celery in
Spain and development of a real-time PCR method for its
detection. Phytopathology 104:804–811.
Tingley, M. W., and S. R. Beissinger. 2009. Detecting range
shifts from historical species occurrences: new perspectives on
old data. Trends in Ecology & Evolution 24:625–633.
Tran, L. T., S. P. Worner, R. J. Hale, and D. A. J. Teulon. 2012.
Estimating development rate and thermal requirements of
Bactericera cockerelli (Hemiptera: Triozidae) reared on
potato and tomato by using linear and nonlinear models.
Environmental Entomology 41:1190–1198.
Turek, D., de Valpine P., and C. J. Paciorek. 2016. Efficient
Markov Chain Monte Carlo sampling for hierarchical hidden
Markov models. arXiv preprint arXiv:1601.02698.
Tuthill, L. D. 1945. Contributions to the knowledge of the
Psyllidae of Mexico (continued). Journal of the Kansas
Entomological Society 18:1–29.
U.S. Environmental Protection Agency (USEPA). 1998. Guide-
lines for ecological risk assessment. U.S. Environmental
Protection Agency, Washington, D.C., USA.
van der Putten, W. H. V., M. Macel, and M. E. Visser. 2010.
Predicting species distribution and abundance responses to
climate change: why it is essential to include biotic interac-
tions across trophic levels. Philosophical Transactions of the
Royal Society B 365:2025–2034.
van Strien, A. J., T. Termaat, D. Groenendijk, V. Mensing, and
M. K
ery. 2010. Site-occupancy models may offer new oppor-
tunities for dragonfly monitoring based on daily species lists.
Basic and Applied Ecology 11:495–503.
van Strien, A. J., C. A. Swaay, and T. Termaat. 2013.
Opportunistic citizen science data of animal species produce
reliable estimates of distribution trends if analysed with occu-
pancy models. Journal of Applied Ecology 50:1450–1458.
van Strien, A. J., D. L. Bekker, M. J. J. La Haye, and T. van der
Meij. 2015. Trends in small mammals derived from owl pellet
data using occupancy modelling. Mammalian Biology 80:
340–346.
Wallis, R. L. 1955. Ecological studies on the potato psyllid as a pest
of potatoes. US Department of Agriculture, Washington, DC.
Wieczorek, J., Q. Guo, and R. Hijmans. 2004. The point-radius
method for georeferencing locality descriptions and calculat-
ing associated uncertainty. International Journal of Geo-
graphical Information Science 18:745–767.
Workneh, F., D. C. Henne, J. A. Goolsby, J. M. Crosslin, S. D.
Whipple, J. D. Bradshaw, A. Rashed, L. Paetzold, R. M.
Harveson, and C. M. Rush. 2013. Characterization of man-
agement and environmental factors associated with regional
variations in potato Zebra Chip occurrence. Phytopathology
103:1235–1242.
SUPPORTING INFORMATION
Additional supporting information may be found online at: http://onlinelibrary.wiley.com/doi/10.1002/eap.1569/full
DATA AVAILABILITY
Data available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.24mk0
September 2017 PEST OUTBREAKS AND MUSEUM SPECIMEN DATA 1837
