Shifting roles of abiotic and biotic regulation of a multi-host parasite following disturbance

18 Pages·2005·0.2 MB·English

10-Collinge-Chap10.qxd 24/12/05 07:54 AM Page 135
CHAPTER 10
Shifting roles of abiotic and biotic
regulation of a multi-host parasite
following disturbance
Mary F.Poteet
10.1 Background disturbance in macroparasitic diseases with
complex life cycles (Wasserberg et al. 2003). For
Impacts of macroparasitic diseases can scale from
example, overfishing in Lake Malawi is hypothes-
the individual host to the community (Anderson
ized to increase densities of the intermediate host
and May 1978; Holmes 1982; Scott 1988; Scott and
snail of schistosomiasis and lead to increased
Dobson 1989), however our knowledge of the
prevalence in humans (Stauffer et al. 1997).
community-level mechanisms that drive macro-
Although this is a compelling argument, the
parasite dynamics in wildlife is still limited. Even
authors acknowledge that their data are anecdotal
more limited is empirical evidence that shows how
and more study is necessary to test their hypothesis.
parasites of wildlife respond to anthropogenic
A recent paper also suggested that eutrophication
disturbances. Many of the current emergent and
could alter food webs in a cascading effect from
resurgent parasitic diseases in wildlife appear to be
snail hosts to parasite abundance to limb deform-
associated with anthropogenic activities (Schrag
ities in amphibians (Johnson and Chase 2004).
and Wiener 1995; Harvell et al. 1999; Daszak et al.
Again, this is a compelling hypothesis that deserves
2000), but the paucity of rigorous data on wildlife
further testing but the data that support the links
diseases in general and their ecological aspects in
between eutrophication, snails, and infection in
particular limit our ability to assess causality. There
amphibians is generated from different studies in
are cases in which causality can be assigned, but
different sites over different years (see Skelly et al.,
these generally are limited to zoonoses (e.g. Lyme
chapter 11, this volume for a systematic study of
disease, Spielman 1994; LoGiudice et al. 2003;
urbanization effects on trematode infection in
Wasserberg et al.2003) and introduced pathogens of
amphibians).
endangered or charismatic species (e.g. malaria in
Due to the extent and severity of anthropogenic
Hawaiian birds (Van Riper et al. 1986), whirling
disturbances and the likelihood that they will
disease in trout (Hedrick et al. 1998), distemper in
significantly alter parasite–host dynamics, it is essen-
black-footed ferrets (Thorne and Williams 1988),
tial to rigorously measure the response of different
upper respiratory disease in desert tortoises
types of parasites and pathogens to disturbances
(Jacobson et al. 1991)). To date, few studies rigor-
and then apply this knowledge in ecological
ously test for the impacts of anthropogenic dis-
restoration and conservation efforts. This chapter
turbances, such as habitat loss, change, and
focuses primarily on macroparasites that are espe-
degradation, on macroparasitic diseases in wildlife.
cially sensitive to disturbance due to the diversity
The evidence needed to link disturbance with
of hosts and transmission stages necessary to
diseases is often anecdotal, even for well-studied
complete their life cycles. In addition to their inherent
human diseases (McSweegan 1996). Recent work
sensitivity to disturbance, macroparasites can have
provides clues to the importance of anthropogenic
135
10-Collinge-Chap10.qxd 24/12/05 07:54 AM Page 136
136 DISEASE ECOLOGY
significant, but difficult to detect effects on their watershed. This eliminates the need to estimate
host populations. Indeed, recent work demon- parasite loss due to host movement. I carefully
strates that nematode parasites can cause popula- selected 18 creeks in watersheds exposed to three
tion cycles in grouse (Hudson et al. 1998). Thus, a levels of disturbance associated with clearcut
change in macroparasitism in response to logging. They include streams located in old growth
disturbance could greatly affect host populations, forests without anthropogenic disturbances,
and the dynamics between hosts and parasites streams in clearcut forests, and streams in clearcut
(May and Anderson 1979). forests that experienced severe winter flooding
One of the main limitations in assessing impacts (debris flow) that was exacerbated by logging. I
of anthropogenic disturbances on parasite popula- measured the effects of disturbance level on the
tions in wildlife is the inability to replicate these density of each obligate host and on the prevalence,
disturbances. Natural disturbances play a strong intensity, and density of each stage of the parasite. I
role in structuring patterns of species diversity and then assessed whether disturbance alters the
distribution (Sousa 1984, 2001; Pickett and White strength or direction of the relationship between
1985). The effects of natural disturbances are often host and parasite densities and explored the
measured through replicated experimental manip- implications of these changes to parasite
ulations within a designated temporal and/or transmission.I interpret these results in the context
spatial regime (Levin and Paine 1974; Sousa 1979; of general Anderson and May (1978, 1979) models
Hobbs and Mooney 1991). The effects of large-scale, for macroparasites. These models assume mass-
single-point anthropogenic disturbances, such as action transmission that results in a positive
oil spills, can be difficult to assess as they are often relationship between host and parasite densities
generated by events that are unreplicated through with or without an upper threshold. These models
time or space (for review see Schmitt and Osenberg are commonly used to predict the response of
1996). As long as care is taken in site selection, some parasites to environmental disturbances (e.g.
anthropogenic disturbances, such as habitat frag- Lafferty and Holt 2003). Finally, I test whether
mentation and road building, can be used for disturbance shifts the role of environmental and
natural experiments; however these are the exception biotic factors in the host–parasite populations.
to the rule. In addition, understanding the effects of
disturbance on epidemiology can be made more
10.2 The parasite-host system
difficult due to dispersal of infected individuals,
parasite stages, or vectors away from or into the site The trematode parasite, Cephalouterina dicamptodoni,
of disturbance. Parasites with complex life cycles obligately and sequentially infects three stream-
often infect hosts that cross ecosystem boundaries, dwelling species: the Pacific giant salamander,
such as schistosomes in humans, or migrate long Dicamptodon tenebrosus, an abundant snail, Juga
distances, such as Nanophyetus in salmon. The silicula, and the stonefly, Calineuria californica (Fig.
difficulty is in assessing not only the patterns of 10.1) (Senger and Macy 1953; Schell 1985). Adult
infection, but also whether and how they are parasites live and sexually reproduce in the intes-
affected by anthropogenic impacts. tines of their salamander hosts. Parasite eggs are
In this chapter, I report results of a study released into the environment with the host’s feces
designed to measure the effects of three levels of where they are consumed by snails and sub-
anthropogenic disturbance on each host and sequently hatch into miracidia. The miracidia
parasite stage of a macroparasite that occurs in migrate to the snail’s gonads and transform into a
stream communities of the Pacific Northwest, USA. series of sporocyst stages. Each of these sporocysts
The disturbance of interest is clearcut logging: a asexually produces thousands of cercariae that are
prevalent, replicated, large-scale disturbance in the then released into the aquatic environment. To con-
northwestern United States. I chose a “closed” tinue the life cycle, these cercariae search for the
parasite–host system in which none of the hosts next obligate host species which is a nymphal
regularly disperses or migrates outside of the local stonefly, C. californica. Once found, the cercaria
10-Collinge-Chap10.qxd 24/12/05 07:54 AM Page 137
SHIFTING ROLES OF ABIOTIC AND BIOTIC REGULATION 137
Dicamptodon
tenebrosus
Predation
Egg
Figure 10.1 Life cycle diagram of Cephalouterina
Adult
dicamptodoni,a trematode parasite that infects the
Calineuria Juga definitive host,Dicamptodon tenebrosus,the first
californica Sporocyst silicula intermediate host,Juga silicula,and the second
intermediate host,Calineuria californica.Free-living
Metacercaria parasite stages transmit to the snail and stonefly
host.Parasites transmit to the salamander via
predation of an infected stonefly.Arrows indicate
direction of transmission for each stage of the
Cercaria parasite between hosts.
pierces the stonefly with a scalpel-like stylet and invertebrate biomass in streams, reaching densities
penetrates the exoskeleton through the incision. of 1500m2, thus making them strong competitors
Cercariae encyst as metacercariae in the muscle with macroinvertebrate grazers, particularly
tissues of the stonefly. The life cycle of the parasite Diptera (Hawkins and Furnish 1987).
is completed when a salamander preys upon an Calineuria californica, the second intermediate
infected stonefly and the metacercaria excysts and host, is distributed from central California north
grows to sexual maturity. through British Columbia and east to Montana
Each host of C. dicamptodoni is an important (Stewart and Stark 1993). In the central Cascades,
component of the aquatic food web in small stonefly nymphs are semivoltine. Calineuria prey
streams in the Cascade Mountains of Oregon. dominantly on larval Diptera and Ephemeroptera
Larvae of D.tenebrosusoften comprise the greatest (Sheldon 1969; Peckarsky 1984).
vertebrate biomass in small streams throughout
western Oregon and California and are voracious
predators (Murphy and Hall 1981; Corn and Bury 10.3 Methods
1989). Dicamptodon larvae eat a wide variety of
10.3.1 Host censuses
aquatic and terrestrial prey, with aquatic macroin-
vertebrates comprising the most significant prey To measure the effects of anthropogenic disturbance
base (Parker 1992). Salamander larvae of different on patterns of parasite and host abundance,
size classes overlap significantly in the prey they I surveyed each host species from 18 streams
consume with preference toward certain macroin- located in the McKenzie and Fall creek watersheds,
vertebrates, including Calineura spp. During their Oregon. The streams were located in small water-
aquatic stage, D. tenebrosus are territorial and can- sheds (4–16 km2) that had experienced one of three
nibalistic. With the exception of neotenic individu- levels of anthropogenic disturbance associated with
als, larval salamanders reside in streams for 2–3 logging. The three categories of disturbance were:
years until metamorphosis to the terrestrial adult no disturbance (ND), intermediate disturbance
stage (Nussbaum and Clothier 1973). (ID), and severe disturbance (SD). ND streams
Juga siliculais an abundant aquatic snail inhabiting (n=7) were located in watersheds covered by
mid-elevation streams from northern California mature forests that were at least 80 years old. ID
north to southern Washington (Burch 1982). This streams (n=7) were in watersheds with at least 50%
species often comprises the majority of total clearcuts. SD streams (n=4) were in watersheds
10-Collinge-Chap10.qxd 24/12/05 07:54 AM Page 138
138 DISEASE ECOLOGY
with predominantly clearcut forests that experienced Table 10.1 Number of D. tenebrosuslarvae captured and dissected
major flooding and debris flows (as defined in along a 150 m transect within each creek.All captured larvae were
weighed and measured in the field.Larvae that were not dissected
Leopold et al. 1964) 2 years prior to the survey.
were released in the creek from where they were captured.
Clearcut watersheds were defined as having
(cid:2) 20-year-old clearcuts covering at least 50% of the Creek Date Disturbance Captured Dissected
area of the watershed. Each watershed was selected
CONE 17 June ND 42 18
to minimize differences in logging-independent geo-
JONES 21 June ND 74 23
morphology such as stream gradient, elevation, and
SLICK 24 June ND 12 10
watershed size. Streams were randomly sampled
SFGATE 28 June SD 24 10
with respect to date from June 13 through July 30, HOLDEN 1 July ID 4 3
1998. OSBORNE 3 July ID 43 20
Each stream was sampled over a 2-day period. RITCHIE 6 July ID 46 21
Macroinvertebrates and physical data were FINN 8 July ID 14 10
collected on the first day and salamanders were WFNFGATE 10 July SD 31 17
collected on the second day. Streams were sampled HAGAN 11 July ND 100 30
EFNFGATE 15 July SD 48 19
in a stratified random manner and samples were
SIMMONDS 17 July ND 38 17
collected every 5 m along a 150 m transect for a
ELK 19 July ND 24 11
total of 30 surber samples per stream. D. tenebrosus
WFDEER 21 July ID 46 23
larvae and neotenes were collected from each
EFDEER 23 July ID 42 19
stream along the 150 m transect using a fish elec-
NFNFGATE 25 July SD 21 11
troshocker. Seine nets were placed at least every MONA 28 July ND 45 22
50 m along the transect to block individuals from STURDY 31 July ID 8 5
escaping. All D. tenebrosuswere counted, measured,
Disturbance levels are ND=no disturbance,ID=intermediate
and weighed in the field. Arepresentative sample
disturbance,SD=severe disturbance.
of larvae was euthanized, dissected (Table 10.1),
and all intestinal parasites were counted and stage averaged across all infected hosts in the
preserved. population. Mean parasite density was calculated
Snails, stoneflies, and other macroinvertebrates as the number of parasites of a single stage
were collected from a 30 (cid:3) 30 cm quadrat using a averaged across all hosts per square meter. For
surber sampler. All snails greater than 5mm were parasites in snails and stoneflies, each metric was
dissected for sporocyst presence. Counting sporo- averaged per quadrat. Stream averages were calcu-
cysts was time-prohibitive so intensity of this stage lated as the grand mean for all quadrat means.
was estimated based on a regression of percentage Means for prevalence, intensity, and density of the
of gonad infected and gonad length (Poteet 2001). adult parasites were averaged across all dissected
Snails smaller than 5mm were not included in salamanders and then divided by total creek area
these data since prior dissections revealed that sampled. The total creek area sampled was the
small snails were never infected (Poteet 2001). All product of the transect distance of 150m and the
encysted metacercariae were counted in each stone- mean wetted width of the creek.
fly. To count all the metacercariae, each stonefly
was digested in an HCl and pepsin bath that dis-
10.3.2 Disturbance,environmental variables,
solved stonefly muscles, but not parasite cysts (Ash
and parasitism
and Orihel 1991).
Prevalence, intensity, and density of each parasite Logging leads to increased stream temperatures,
stage were calculated following Bush et al. (1997). altered stream velocity, and decreased habitat
Prevalence was calculated as the proportion of heterogeneity. Clearcuts in the Cascade mountains
hosts in the population that were infected by one are associated with high rates of hillslope erosion
parasite stage. Mean parasite intensity was and increased storm peak discharge (Jones and
measured as the number of parasites of a single Grant 1996) that affect stream channel morphology,
10-Collinge-Chap10.qxd 24/12/05 07:54 AM Page 139
SHIFTING ROLES OF ABIOTIC AND BIOTIC REGULATION 139
particle size, and discharge. Each of these variables variables were transformed into their principal
can significantly affect the success of parasite components (Selvin 1998; S-Plus 2000). Variables
transmission (Chernin and Perlstein 1969; Upatham used in the PCAincluded maximum water temper-
1974; Anderson et al. 1982; Jewsbury 1985; ature, thalweg velocity, pool/riffle ratio, percentage
Woolhouse and Chandiwana 1989; Shostak and canopy cover, channel shape (measured as width/
Esch 1990; Sousa and Grosholz 1991; McKindsey depth ratio), embeddedness (Brusven Index,
and McLaughlin 1994). Along each 150m transect, Gordon et al.1992), and mean substrate particle size
I measured a series of environmental attributes (Dunne and Leopold 1978). These variables were
associated with logging and used these measures to chosen for their high probability of affecting host or
correlate parasitism with disturbance-related envi- parasite populations. Before conducting the ana-
ronmental variables. These attributes included lysis, the data were transformed where necessary to
dominant substrate particle size, thalweg depth, meet the assumptions of normality and homogen-
thalweg velocity, percent channel morphology eity of variance and all variables were standardized
(pool, riffle, run, or cascade), and percentage of to a mean of 0 and unit variance to account for
canopy cover. I also measured substrate embed- differing scales of measurement.
dedness, or the extent to which fine sediments Biotic variables considered for the ANCOVA
filled the interstitial spaces around larger substrates included host length, host density, parasite density
(Bovee 1982; Gordon et al. 1992). The interstitial and, in the case of metacercarial transmission,
spaces formed within creek substrates are impor- salamander diet (i.e. the mean number of stoneflies
tant habitat for salamanders, stoneflies, and other found in salamander stomachs per treatment, see
stream biota. Thus, as embeddedness increases, habi- Fig. 10.1). Abiotic variables considered for the
tat availabilityand biotic productivity decreases. ANCOVA included the first three principal
To standardize host and parasite density components. Because the number of possible
measurements across creeks, I measured bank-full explanatory variables for the ANCOVAwas greater
width and depth and wetted-channel width. Air than the number of replicate streams, I chose leaps
temperature was measured in the morning, and and bounds analyses to select the best-fit models
minimum and maximum diurnal water tem- for each host–parasite pair (Furnival and Wilson
peratures were measured over the 48-h sampling 1974; S-Plus 2000). Leaps and bounds analysis is
period. Watershed area was calculated from USGS similar to model selection analyses that evaluate
7.5’ topographic maps. the best fit models from the full set of explanatory
variables. However, unlike other model selection
analyses leaps and bounds uses an algorithm
10.3.3 Statistical analyses
developed by Furnival and Wilson (1974) that
I tested for differences in host density and parasite identifies the best fit by testing all possible models.
prevalence, intensity and density across forest The resulting best-fit models of donor-target pairs
disturbance levels with analysis of variance had at least four explanatory variables and thus I
(ANOVA) followed by adjusted Tukey post hoc was unable to divide the forests into all three dis-
comparisons. I regressed parasite density on host turbance levels for the analyses due to low replica-
density for each host–parasite pair to test model tion. The final best-fit models lumped ID with SD
assumptions that parasite transmission leads to into a single “disturbed” category.
positive correlation between host and parasite
densities (May and Anderson 1979). All variables were 10.4 Results
transformed where necessary to meet the assump-
10.4.1 Responses of host density and parasite
tions of homogeneity of variance and normality.
prevalence,intensity,and density to
I tested for the responses of parasite density to
environmental disturbance
abiotic and biotic variables with principal com-
ponents analysis (PCA) followed by analysis of Densities of each host species were significantly
covariance (ANCOVA). Seven correlated abiotic lower in logged creeks, but patterns differed by
10-Collinge-Chap10.qxd 24/12/05 07:54 AM Page 140
140 DISEASE ECOLOGY
(a) Salamander Snail Stonefly
0.12 250 30
a c
a 200 a
y
nsit 0.08 150 b 20 a
e
d b b
ost 0.04 100 10 b
H
50
0 0 0
(b) Adult parasite Sporocyst Metacercaria
100 10 100 a a
ns 8 a
75 75
e b
nc 6 a
e
al 50 50
ev 4
r
P
25 2 b 25
0 0 0
Figure 10.2 Host density,parasite prevalence,and
(c) Adult parasite Sporocyst Metacercaria
parasite intensity for 18 streams of three disturbance
8 1600 a 8 intensities.(A) Host density is the mean number of
a,b a hosts/m2.(B) Prevalence is the mean proportion of
6 a 1200 6 hosts infected.(C) Parasite intensity is the mean
sity ns number of parasites averaged across all infected
en 4 800 b 4 hosts.Small letters denote statistical significance at
nt the p< 0.05 level.Error bars are±1 SE.
I b
2 400 2 White bars = no disturbance (ND) sites,
grey bars = intermediate disturbance (ID) sites,and
black bars = severe disturbance (SD) sites.
0 0 0
host species and level of disturbance (Fig. 10.2(a)). (Fig. 10.3(a)). These results run counter to many
Salamander density was similar across ID and SD population models of macroparasite population
creeks, but snail densities were lower only in creeks dynamics (e.g. May and Anderson 1979) and to
with the most severe disturbances. Stonefly predictions that parasite densities will decrease
densities differed across disturbance levels such following disturbances that cause decreased host
that the lowest stonefly densities were found in ID densities (Lafferty and Holt 2003). This anomaly is
streams and the highest densities were found in SD likely explained by an increase in the rate of
streams. parasite transmission from stoneflies to salaman-
Prevalence and intensity of each parasite stage ders that result from the predation-dependent
was generally lower in logged sites but did not transmission of this parasite stage (Box 10.1).
track host density (Fig. 10.2(b) and (c)). Interestingly, Of all stages, the sporocysts were most affected
adult parasite intensity recovered somewhat in by disturbance, with significant declines in
severely disturbed streams even though salaman- prevalence and intensity in ID and SD creeks. This
der densities remained low (Fig. 10.2(a) and (c)). At density-dependent effect is partially explained by the
the same time, prevalence of adult parasites sharp decline in snail host density in the sites with
remained constant across disturbance (Fig. 10.2(b)) severe disturbance, but is also associated with
leading to a recovery in the density of adult changing abiotic conditions (see Principal Components
parasites in SD streams as compared to ID streams Analyses, below). Intensity of metacercariae was not
10-Collinge-Chap10.qxd 24/12/05 07:54 AM Page 141
SHIFTING ROLES OF ABIOTIC AND BIOTIC REGULATION 141
(a) 0.8 low recruitment of cercariae from small snail
populations at these severely disturbed sites.
2m)
/
ult (# 0.4 1as0s.4u.m2 pEtniovnirsonmental disturbance and model
d
A
If the assumption of mass-action transmission is
0 correct in the general macroparasite models (May
0 0.06 0.12 0.18 and Anderson 1979), then regressions of parasite
Salamander (#/m2) density on host density should have positive
(b) 1500 slopes. Significantly positive regressions between
parasite and host densities were found in only two
2m) cases (Fig. 10.3(a) and (c)): between salamanders
/1000
st (# and adult parasites in ID (R2=0.5584, F1,5=6.321,
y p=0.054) and between stoneflies and metacercariae
c
ro 500 inND (R2=0.6153, F =7.999, p=0.037). In general,
po 1,5
S the regression slope between adult parasites and
0 salamanders trended toward positive across all dis-
0 100 200 300 400 turbance levels (Box 10.1). This is not the case for
Snail (#/m2) the two other parasite stages. In fact, the regression
(c) 600 between snails and sporocyst densities was signi-
2m) ficantly negative across ID sites (R2=0.6096,
F =7.806, p=0.038) and was not significantly
/ 400 1,5
#
a ( different from 0 at ND and SD sites (Fig. 10.3(b)).
ri
a
rc 200
e
c 10.4.3 Physical attributes of streams:univariate
a
Met analyses
0
0 10 20 30 When selecting streams to test for effects of
Stonefly (#/m2) environmental disturbance on parasite populations,
I controlled for geographic location, watershed
area, elevation, and stream gradient to minimize
Figure 10.3 Regressions between each host and parasite pair at
differences among watersheds that were not caused
three levels of disturbance.Common models for macroparasite
by disturbance. All but three streams were located
populations assume a positive correlation between parasite and host
densities.These regressions demonstrate that this assumption is not in the McKenzie River watershed. The remaining
valid for most of the parasite stages in areas with environmental three streams, Jones, Slick, and Sturdy, were in
disturbances.(a) Adult parasite in salamanders,(b) sporocyts in snails, the Little Fall creek watershed, which is the
(c) metacercariae in stoneflies.Open triangles with dashed lines = no
drainage immediately south of the McKenzie
disturbance (ND) sites,gray squares with solid gray lines = intermediate
River watershed. There were no differences in
disturbance (ID) sites,and filled circles with solid black lines = severe
disturbance (SD) sites. watershed area, elevation or stream gradient
among streams in different forest types (basin
statistically different across the disturbance gradient area: F =0.271, p=0.766; elevation: F =1.743,
2,15 2,15
(Fig. 10.2(c)), but metacercarial density was lower p=0.208; gradient: F =0.052, p=0.949).
2,15
in SD than ND sites (Fig. 10.3(c)). The low meta- Undisturbed streams had significantly higher
cercarial density in SD sites was somewhat surpris- pool to riffle ratios (30%) than ID (11%) or SD (2%)
ing since stonefly densities in these sites were the streams. Thalweg velocity and substrate embed-
highest of the three stream types. The low density dedness (Brusven Index) were higher in streams
of metacercariae in SD sites was probably due to at logged sites (velocity: F =28.276, p=<0.001;
2,15
10-Collinge-Chap10.qxd 24/12/05 07:54 AM Page 142
142 DISEASE ECOLOGY
Box 10.1 Transmission through predation of an obligate,intermediate host
Predicting how changes in salamander density will affect by the functional response of the predator,but also by the
parasite dynamics is not straightforward since salamanders proportion of stoneflies that are infected.Thus,a
acquire infections by consuming infected stoneflies.Thus, transmission term that is dependent upon prevalence of
transmission of metacercariae will depend upon the infection in stoneflies would provide a better description of
predator–prey functional response curve between metacercarial transmission than would mass-action
salamanders and stoneflies.May and Anderson (1979) transmission.This is reminiscent of frequency-dependent
suggest that transmission completed through a trophic link models in which infection of the vector is dependent upon
is extremely efficient and that the threshold density of the frequency of infected nonvector hosts (Getz and
predators,in this case salamanders,required to maintain Pickering 1983;Thrall et al.1995;McCallum et al.2001).
the adult parasite population is low (May and Anderson One interesting outcome of frequency-dependent
1979).So even at extremely low salamander densities, transmission is the lack of a host threshold,since
metacercarial transmission should be successful.However, transmission depends upon proportional infection.By this
the rate of transmission will vary with the predator–prey model,even if stonefly and salamander densities are
functional response,which will depend upon the density of extremely low,parasites should still be able to establish in
salamanders and stoneflies.Pacific giant salamander larvae salamanders.That seems to be the case in this system,
prey disproportionately on mayflies and large predaceous where even at high levels of disturbance and low
insects,including stoneflies (Parker 1994).The proportion of metacercarial densities (Fig.10.3(c)) adult parasite density
stoneflies relative to the other prey items preferred by almost recovers to undisturbed levels (Fig.10.3(a)).In
salamanders is not constant across forest type (M.Poteet, either transmission model,decreased salamander density
unpublished data).This suggests that the functional should result in decreased metacercarial transmission and
response curve is likely to be nonlinear with disturbance, this is the case here.At each level of disturbance,
either Type II or Type III,which would lead to nonlinear regardless of salamander density,there is a direct
parasite transmission and not the linear transmission correlation between salamander densities and adult
function assumed by mass-action models (e.g.Crofton parasites.This is not the case for the free-living
1971;Anderson and May 1978).In addition,since the transmission stages (sporocysts and metacercariae),whose
salamander must eat an infected stonefly to become densities are affected by both environmental stressors and
infected,the rate of transmission will not only be affected host density.
embeddedness: F =5.391, p=0.005). The only changes altered channel shape, increased stream
2,15
abioticfactor that was significantly lower at logged velocity, and decreased streambed particle size.
sites was canopy cover (F =20.770, p<0.001). Each of these features is described by at least one
2,15
The decreased canopy cover in ID and SD streams principal component, with channel shape a
corresponded to slightly increased air temperature significant feature in each principal component.
and maximum water temperature, though The first principal component (PC1) described
neitherdiffered significantly with disturbance level channel shape and was composed of positive
(air temperature: F =2.667, p=0.102; water correlations between stream velocity and the ratio
2,15
temperature:F =0.648, p=0.537). of bank full width to bank full depth (width/
2,15
depth). This reflected higher stream velocities and
wider, shallower channels following logging. The
10.4.4 Physical attributes of streams:principal
second principal component (PC2) described
components analyses
stream channel pool/riffle morphology. Logging
Each of the principal components described some increased the proportion of riffle habitat which led
aspect of the physical responses of streams to log- to greater mean channel width. Pool/riffle ratio
ging. Logging increased sediment load delivered to was inversely proportional to width/depth ratio
the creeks, increased peak storm flows (Jones and and stream temperature. The third principal
Grant 1996), and decreased canopy cover. These component (PC3) described streambed particle size
10-Collinge-Chap10.qxd 24/12/05 07:54 AM Page 143
SHIFTING ROLES OF ABIOTIC AND BIOTIC REGULATION 143
33669354
65075070
61031500 o
Table 10.2Results of ANCOVA testing for the effects of disturbance on the biotic and abiotic variables that best explain variation in each stage in the life cycle of the parasite.Best fit modelswere selected with the leaps and bounds method (see text for explanation) Adult SporocystMetacercaria CtpCtpCtp ND–0.504–0.9490.361ND–31.663–2.250.059ND–0.987–0.4550.D–0.469–0.8830.394D–20.61–1.4640.186D–3.476–1.6010.222Salamander #/m(ND)0.2362.2240.046Snail #/m(ND)8.7972.7260.03Stonefly #/m(ND)3.4073.9070.222Salamander #/m(D)0.1793.5540.004Snail #/m(D)0.9051.2240.26Stonefly #/m(D)0.8580.9460.Salamander SVL (ND)0.2741.9120.080PC3 (ND)–4.02–1.2670.246PC1 (ND)1.4681.5780.Salamander SVL (D)0.1132.5750.024PC3 (D)0.4940.8310.434PC1 (D)–0.217–0.7050.2PC1 (ND)7.5252.4640.043Snail #/m(ND)2.8282.0860.2PC1 (D)–0.946–2.9430.022Snail #/m(D)1.744.1090. ND=no disturbance,or watersheds in old growth forests;D=disturbance,or watersheds in clearcut forests with and without debris flows;SVL=salamander length measured from the snout tthe vent in mm;P1=principal component 1;P3=principal component 3;C=regression coefficients.The first two coefficients in the ‘C’column are regression intercepts for the ND or D sitesrespectively.The remaining coefficients are regression slopes for each variable for the ND or D sites respectively.
10-Collinge-Chap10.qxd 24/12/05 07:54 AM Page 144
144 DISEASE ECOLOGY
as a function of channel shape and was composed Unlike adult parasites, disturbance caused a
of the pebble count, maximum water temperature, significant shift in the type of variables that con-
and width/depth ratio. Width to depth ratio and trolled sporocyst density. Both abiotic and biotic
pebble count were positively correlated with each variables correlated with sporocyst density in
other and inversely related to the maximum water unlogged streams, but only abiotic variables
temperature. The first three principal components affected sporocyst density in logged streams.
among all streams explained 76% of the variation in Specifically, sporocyst density was significantly
abiotic variables. The significant components were and positivelycorrelated with snail density and the
based on variables with loadings greater than 0.40 channel shape/velocity principal component
or less than (cid:4) 0.40 (Hair et al.1987; McGarigal et al. (PC1). However, in streams with disturbance,
2000). sporocyst density was significantly and negatively
correlated with PC1 and not correlated with snail
density (Table 10.2).
10.4.5 Disturbance shifts the importance
Disturbance caused a shift in the biotic variables
of biotic and abiotic variables
controlling metacercarial densities, but as with
Because host density did not explain significant adult parasite density, abiotic variables were not
amounts of variability in abundance for most of associated with metacercarial densities in logged or
the parasite stages, I expanded the regression unlogged streams. In unlogged streams, stonefly
analyses to include host density and the first density explained the majority of variation in
three principal components. Leaps and bounds metacercarial density. Disturbance shifted control
analyses selected the following explanatory over metacercarial density from stonefly hosts to
variables for each parasite stage to minimize the snail hosts in logged streams (Table 10.2).
Mallow’s Cp:
(cid:2) adult parasite density: salamander density and 10.5 Discussion
salamander length;
Land use change and resource use by humans
(cid:2) sporocyst density: snail density, PC3, and PC1;
increasingly force species to cope with fragmented
(cid:2) metacercarial density: stonefly density, PC1, and
and degraded habitats that can alter infection
snail density.
dynamics of diseases and parasites (e.g. Dobson
The models were analyzed with ANCOVA and Carper 1992; Lafferty and Holt 2003; LoGiudice
using forest type as a covariate. Since there were et al. 2003, this volume). Although several reviews
only four replicates of SD streams, I combined report that diseases and parasites of wildlife
the ID and SD forests into a “disturbed” category respond to anthropogenic disturbances (Schrag and
(D). Wiener 1995; Lafferty 1997; Daszak et al.2000), most
In all but the adult parasite stage, disturbance of the studies cited are either anecdotal or are not
switched the dominant variable that best explained sufficiently replicated, which limits our ability to
parasite density (Table 10.2). Although disturb- interpret them (but see Wasserberg et al. 2003). By
ances had significant effects on the overall density measuring each stage of a macroparasite and its
of adult parasites, they did not alter the positive host species across replicated environmental
correlation between the parasite and its salamander disturbances, this study measures the response of
host. For example, adult parasite density was parasite infection levels to disturbance and also
positively and significantly correlated with sala- explores the mechanisms that drive these changes.
mander density in both logged and unlogged Disturbances caused by logging led to significant
streams, and significantly correlated with salaman- declines in host density and in the prevalence,
der size in logged streams (Table 10.2). Adult intensity, and density of infection by trematodes
parasite densities could not be explained by in these three host species. Regressions between
variation in abiotic variables for either logged or parasite and host densities were also affected
unlogged streams. by disturbance, suggesting that disturbance