195

Comparative effects of mites and lice on the reproductive

success of rock doves (Columba livid)

D. H. CLAYTON and D. M. TOMPKINS

Department of Zoology, University of Oxford, South Parks Road, Oxford OX1 3PS, UK

(Received 15 March 1994; revised 19 July 1994 ; accepted 19 July 1994)

SUMMARY
We report experimental data comparing the effects of Mesostigmatid mites and Ischnoceran lice on the reproductive
performance of a single group of captive rock doves (Columba livia). Several components of host reproductive success were
compared for the two groups, including number of eggs laid, hatching success, nestling growth rates, fledging success,
post-fledging body mass and survival. Adult body mass and survival were also compared. There was a dramatic difference
in the effects of the mites and lice. The former drove host reproductive success to zero, mainly by agitating adults and
causing them to incubate eggs less faithfully. Nestling growth rates and post-fledging survival were also significantly
reduced by mites. Lice showed no effect on reproductive success whatsoever, even though the feather damage they cause
is known to have energetic consequences (Booth, Clayton & Block, 1993). Neither parasite had a significant effect on adult
birds. Although Ischnocera are found on most species of birds, our results for lice constitute the first experimental test
of the impact of Ischnocera on avian reproductive success (preliminary report by Clayton & Tompkins, 1994). We discuss
reasons for the different effects of mites and lice, including the relationship of horizontal (mites) and vertical (lice)
transmission to the evolution of virulence.

Key words: ectoparasites, reproduction, rock doves, mites, lice.

the effects of mites and lice over a longer-term

INTRODUCTION experiment, including data for a larger number of
Theory suggests that the evolution of parasite components of reproductive success, as well as data
virulence is associated with the ease of transmission on adult body mass and survival. We then discuss
among hosts (Anderson & May, 1982; Ewald, 1983). probable constraints on the evolution of virulence in
Parasites transmitted to new hosts independently of Ischnoceran lice.
host fitness (e.g. horizontally transmitted parasites) Experimental tests of the effects of Ischnocera on
may not suffer a reduction in fitness by harming the avian reproduction had not been published prior to
host. However, parasites transmitted vertically from our preliminary report (Clayton & Tompkins, 1994)
parents to offspring will suffer a severe reduction in despite the fact that Ischnocera parasitize most
fitness if they have too great an impact on host species of birds (Marshall, 1981 ; Clayton, Gregory
reproduction. In a recent paper, Lehmann (1991) & Price, 1992). Derylo (1974a, b) reported a re-
compared the effects of dissimilar ectoparasites on a duction in the egg production of Leghorn chickens
population of small mammals (Gerbillus andersoni). parasitized by Ischnocera, but the subjects of study
He showed that independently transmitted fleas also had Amblyceran lice. In contrast to the feather-
reduce host fitness, whereas vertically transmitted feeding Ischnocera, Amblycera feed on the skin and
lice have no detectable effect on components of blood of the host as well as feathers and are known to
fitness. Although the impact of fleas was tested via promote dermatitis and scratching. They are re-
experimental manipulation of flea loads, the results sponsible for up to a 46 % reduction in the egg
for lice were correlational. production of poultry (DeVaney, 1976). The pigeons
We have conducted a parallel study with rock in our study were host only to Ischnocera.
doves (feral pigeons, Columba livia) in which we
compared the effects of independently transmitted
Background
Mesostigmatid mites with the effects of vertically
transmitted Ischnoceran lice. In a preliminary report Rock doves are a monogamous species with life-long
(Clayton & Tompkins, 1994) we showed that mites pair bonds (Levi, 1957; Johnston, 1992). Two eggs
are more virulent than lice, which accords with are normally laid and both parents incubate until the
theory. A survey of the literature provided additional eggs hatch after about 17 days in the nest. T h e
support for the hypothesis that ectoparasite virulence nestlings spend about 1 month in the nest where they
is linked to mode of transmission (Clayton & are fed by a regurgitated mixture of crop milk and
Tompkins, 1994). In the current paper we compare grain by both parents. Feeding by the parents

Parasitology (1995), 110, 195-206 Copyright © 1995 Cambridge University Press

D. H. Clayton and D. M. Tompkins 196

continues for approximately 1 week after the young exposed to fresh air and sunlight (ambient tem-
fledge. Thus, the period of parental care, from laying perature and photoperiod), while being protected
of eggs to independence of fledglings, ranges from 50 from direct wind and precipitation. They were fed
to 60 days. North American rock doves may attempt on a commercial grain mixture (Kaytee Breeder and
to breed year round (Johnston, 1992), but are Conditioner Pigeon Feed #6®) and provided with
normally successful only during March-November grit, nesting material and water (heated during
in Illinois, the site of this study (Clayton, 1989). winter).
The species of mite we studied, Dermanyssus The pigeons nested year round in contiguous,
gallinae, is a member of the family Dermanyssidae open-face boxes mounted on the walls of the
(suborder Mesostigmata). It is a common parasite of enclosure as illustrated in Levi (1957). Most pairs of
both wild and domestic birds (Strandtmann & adults defended two adjacent boxes, enabling them
Wharton, 1958) and is frequently observed in the to overlap clutches with one adult incubating eggs
nests of pigeons (Levi, 1957; Johnston, 1992). D. while the other fed nestlings in the adjacent box.
gallinae spends most of its time in the nest material Eggs were incubated for a mean of 171 days
but emerges periodically to feed on the blood of its (S.D. = 058, N = 55 nests) and the mean number of
host (Sikes & Chamberlain, 1954). It usually feeds post-hatching days in the nest was 30-7 (S.D. = 1-73,
on nestlings, but will also attach to the lower belly of iV=55) (data for April-September 1986). We
adults that are incubating eggs or brooding the periodically culled fledglings to prevent overcrowd-
young. The life-cycle is completed in just 8-9 days ing in the enclosure. The flock contained from one
and consists of the following stages: egg, larva, 2 dozen to several dozen fledglings on any given date.
nymphal instars, and adult (Benbrook, 1965). Adults In addition to the breeders the flock contained
can survive for months without feeding. about a dozen non-breeding 'floaters', resulting in
The species of lice we studied, Columbicola persistent competition for mates and nest boxes.
columbae and Campanulotes bidentatus, are members Fighting over nest boxes was common, although
of the suborder Ischnocera of the insect order residents were seldom usurped. Fighting over nest
Phthiraptera (formerly Mallophaga). They are com- sites is common in free-ranging populations of rock
mon on feral pigeons and are the only Ischnocera doves (D. H. Clayton, personal observation).
found on this species of host. Ischnocera are obligate
parasites that complete their entire life-cycle on the
Counting parasite loads
body of the host (Marshall, 1981). They do not leave
the host's body except to pass among hosts during We use parasite 'load' as a generic phrase en-
periods of direct contact, e.g. between parent birds compassing two explicit measures of parasite abun-
and their offspring. The life-cycle, which requires dance: 'prevalence', the proportion of individual
3—4 weeks, includes an egg, 3 nymphal instars, and birds or nests parasitized; and 'intensity', the
the adult. Eggs are glued to the host's feathers with number of parasites harboured by an individual bird
a glandular cement. C. columbae and C. bidentatus (Margolis et al. 1982).
feed exclusively on the barbules of feathers (Nelson
& Murray, 1971), which they metabolize with the aid Mites. It was not possible to count the total number
of symbiotic bacteria (Eichler et al. 1972; Marshall, of mites in nests non-destructively. Rather, mite
1981). In this paper 'louse load' refers to the intensity was estimated by counting the number of
combined populations of the two species of lice. We mites on nestlings during a careful examination of all
have not distinguished between them as they were body regions. This required less than 1 min for
not manipulated independently and are known to young nestlings but up to 3 min for pre-fledglings.
have similar proximal effects on the host (D. H. Mite counts using this method are significantly
Clayton, personal observation). correlated with the total number of mites in nests
(Clayton & Tompkins, 1994). Data for nestmates
were averaged to avoid pseudoreplication (Hurlbert,
MATERIALS AND METHODS 1984). Nestlings were censused for mites about once
a week between 11 April and 16 June 1987.
Study population
All data were collected between March 1986 and Lice. The louse loads of both adult and nestling birds
March 1988 in an outdoor, free-flight enclosure near were estimated. Nestling louse loads were estimated
the University of Illinois, Urbana-Champaign. The as described above for mites. This approach could
enclosure (9-2 x 37 x 2-5 m) was constructed in 1982 not be used to estimate the louse loads of adult birds
in the loft of a barn and was stocked with 72 feral as their dense plumage made it difficult to see most
pigeons captured on rooftops and under bridges of the lice. Adult loads were therefore estimated on
(Droge, 1986). During our study the flock contained the basis of timed counts of the number of lice on
46 adult breeding pairs from the original stock or various body regions, as described by Clayton
their descendants. The birds were continuously (1991). These counts were substituted into re-
Effects of ectoparasites on reproduction 197

gression models that accurately predict total louse create a 10-30 mm gap between the mandibles
load (r2 ^ 0'82). During this procedure the treatment which prevents the full occlusion of the bill necessary
status of each bird was unknown to the person for efficient preening.
collecting the data. Adults were censused for lice in Since pigeons feed their young by regurgitation,
May and July 1987 and again from November 1987 rather than by forceps-action of the bill, bits do not
to the end of the study in January 1988. interfere with the feeding of offspring. Therefore,
bits have no detectable side-effects on feeding ability
(see below). However, bits might conceivably in-
Manipulating parasite loads
terfere with the ability of birds to defend nest sites
Mites were not present in the enclosure in 1986. since pigeons routinely bite opponents when fighting.
They were first detected on 9 April 1987. The source To control for this possibility, all birds in the captive
of mites is unknown but, as D. gallinae is a generalist flock were bitted at the start of the louse experiment.
occurring on many species other than pigeons, it is Unless fumigated, bitted birds experience a dra-
possible that they were transmitted to the captive matic increase in louse load that mimics increases on
flock by house sparrows {Passer domesticus) nesting birds with natural minor bill deformities (3 of 150
in the barn. wild-caught birds censused for lice in spring 1988
Mites were first censused on 11 April and repeat (Clayton, 1989)). Bit-induced increases in louse load
censuses followed at about weekly intervals through- do not exceed the range of natural louse loads.
out the study. Mites were manipulated immediately All adults in the flock were bitted on 4—6 July
after the second census (17 April) by fumigating 11 1987. On 5-7 August, after ranking birds on the
'low-load' nests, all of which were mite-free during basis of 1986 reproductive success, 1 of every 2 pairs
the first two censuses. The fumigant was a 1 0 % was randomly fumigated (low-load pairs) and the
aqueous solution of pyrethrum. Eleven more 'high- other pair was sham-fumigated with water (high-
load' nests, all of which contained mites during the load pairs). In a few cases randomly assigned
first two censuses, were sham-fumigated with water. treatments were reversed to balance the distribution
(Seven nests used in preliminary trials were excluded of treatments with regard to (1) previous mite
from the experiment.) infestation and (2) physical position of nest boxes. In
Boxes had to be refumigated periodically to total, there were 20 high-load pairs and 23 low-load
prevent increases in mite intensity. Pyrethrum has pairs. Treatments were repeated on 10 September
no side-effects on birds (Jackson, 1985). To be sure and 27 October. At the end of the experiment all
of this, however, we fumigated 10 recently hatched birds had their bits removed.
nestlings and sham-fumigated another 10 with water.
We then compared the growth rates, fledging success
Measuring host fitness
and post-fledging body mass and survival of the two
groups. In 1986, data were collected to establish a baseline
It was possible to control the nest-based mite for reproductive success in the enclosure and to
populations independently of lice, which were con- document seasonal trends. The mite and louse
fined to birds' bodies. This was done by excluding experiments were conducted in 1987. Main analyses
birds from nests for 1-2 hours during fumigation for each experiment were limited to data collected
and sham-fumigation. They were allowed to return over a period of 1 month, beginning 3 days after
to the nests after the fumigant was deactivated by fumigation (20 April-20 May for mites; 10 August-
drying nests with a hair-dryer. 10 September for lice). In the case of the louse
By mid-June rapidly increasing populations of experiment, additional analyses were performed
mites in the enclosure threatened the long-term using data collected from a smaller number of adult
stability of the breeding flock. On 20 June all nest pairs during a follow-up period of 2 months
boxes were fumigated, thus terminating the mite (10 September-10 November). A smaller number of
experiment. Boxes were refumigated as needed over adults had to be used because half of the breeding
the course of the month to eradicate mites from the pairs were removed from the enclosure in late
enclosure. October for use in a mate choice experiment
Lice were manipulated using a two-step pro- (Clayton, 1990). All were returned to the enclosure
cedure. First, preening ability, which is the major in mid-January 1988 and the study was terminated
defence of birds against ectoparasites (Marshall, in March 1988 (see below). The following com-
1981; Clayton, 1991), was impaired in order to ponents of host fitness were measured during the
convert the clumped distribution of lice among birds study.
into a more continuous distribution. Preening was
impaired with ' bits', small (< 0-8 g) C-shaped pieces Egg laying and hatching success. Nests and their
of plated steel inserted between the mandibles of the contents were checked daily. Eggs that failed to
bill and crimped slightly in the nostrils to prevent hatch were discarded several days after the expected
dislodging (but not enough to pierce the tissue). Bits hatching date. The impact of parasites on hatching
D. H. Clayton and D. M. Tompkins 198

was analysed independently by calculating hatching 1-0 -,

success contingent upon the number of eggs laid
•a
during the periods already indicated. 0-9-

Nestling growth and fledging success. Since the mite - 0-8 H

experiment had to be terminated on 20 June, analysis
c
of the impact of mites on growth and fledging 0-7-
success was for nestlings hatched during the month
after fumigation. Fledging was analysed indepen- P 0-6-
dently of pre-fledging components by calculating
fledging success contingent upon the number of eggs 0-5-
hatched. Date of fledging was scored as the first day
a bird was absent from its nest. 0-4
3 4 5
Nestlings were weighed every other day. Some Census
were weighed on the day of hatching, and on odd
days thereafter, whereas others were not weighed
until the day after hatching, and on even days B
thereafter. To make data comparable a fifth-order
-
polynomial regression was fitted to 'even-day'
weights and used to estimate ' odd-day' weights. The 8-
regression did not give accurate estimates for Day 1 >
o
c
so analyses were restricted to Day 3-Day 25 weights. 03
Birds sometimes fledged as early as Day 26; growth CT

had reached an asymptote by this age in any case. All 4-

data for siblings were averaged to provide mean
values per nest. Growth rates were compared using
the general linear model procedure (GLM, con-
trolled for repeated measures) of the Minitab 8.1
Accelerated statistics package (Minitab Inc.).
o-L mflflf]
10
in fn
20 30
n 40
Mite intensity

Post-fledging weight and survival. Fledglings were

12 -i
weighed every 2 weeks for 6 weeks after fledging. It
was not possible to collect data for a longer post-
fledging period because of the need to cull birds to
prevent crowding. Analyses were restricted to birds 8 -
fledged during the periods already indicated. This
c
meant that all young in the mite experiment had 0)
3
either fledged or died in the nest by 20 June, when all O"

nests were fumigated. 4-

Adult weight and survival. Adults were weighed

within 24 h of laying of the second egg of each
clutch. A record of deaths was also kept. In late —I
January the window coverings along one side of the 40 80 120 160
Mite intensity
enclosure were removed, enabling birds to come and
go at will. Although still provided with an unlimited Fig. 1. (A) Increasing mite prevalence over the course of
supply of heated water, the food supply was reduced the study. Mites were censused at weekly intervals
by approximately 50%, thus encouraging birds to beginning 11 April; (B) frequency distribution of mites
forage at other sites such as nearby grain silos. The prior to fumigation in mid-April (iV = 22 nests); (C)
distribution of mites after fumigation. ( • ) Fumigated
survival and body mass of birds still roosting in the
low-load nests; ( • ) sham-fumigated high-load nests.
enclosure were monitored through March 1988 when
Values are the mean number of mites/nest on nestlings
the studv was terminated. 55 15 days of age averaged over weekly censuses from
late April through mid-June.
RESULTS
Prevalence decreased in June owing to a sharp
Parasite loads and transmission
reduction in the occupancy of infested nests by
Mite loads. Mite prevalence increased rapidly from young birds. The earlier increase in prevalence was
mid-April through to the end of May (Fig. 1A). due to widespread horizontal transmission of mites
Effects of ectoparasites on reproduction 199

among nest boxes (see Clayton & Tompkins, 1994). fledged from eggs laid August—September (N — 32
When treatments were assigned (on the basis of the pairs; Wilcoxon T = 81-5, P = 0-58). (Post-fledging
first two censuses), 50% of occupied nest boxes had survival was not measured in 1986.) This similarity
mites (Fig. IB). Although fumigation did not keep is not surprising since the ambient temperatures at
the low-load nests mite-free, it did keep the dis- these times of year were also quite similar (April-
tribution of mites similar to that at the start of the June mean = 23 °C, S.D. = 6 4 3 ; Aug-Oct mean =
experiment (Fig. 1C). Over the course of the 22 °C, S.D. = 6-57; t = 0-70, P = 0-48). Hence there
experiment (results of 8 censuses averaged) low-load was no seasonal variation in host reproductive
nests had a mean intensity of 33 mites (S.D. = 420) success that could have confounded the comparison
compared to a mean of 427 mites (S.D. = 33-42) in of mites and lice.
high-load nests (U = 3, P = 00002). During the louse experiment all adult birds in the
enclosure wore bits. Nevertheless, to assess potential
Louse loads. In May, prior to the louse experiment,
side-effects of bits, the reproductive success of bitted
louse prevalence was 100% and mean intensity was
birds during the louse study was compared to that of
539 lice/bird (S.D. = 1023, N = 86). The lice were
unbitted birds at other times. Two comparisons
overdispersed among host individuals (Fig. 2A),
were made. First, the number of young fledged by
with a distribution that did not differ significantly
unbitted pairs in 1986 did not differ significantly
from a negative binomial distribution (& = l-02;
from the number fledged by the same pairs after
X2 = 2-46, D.F. = 3, P = 0-52 (fitting method of Bliss,
bitting and fumigation in 1987. T h e mean fledging
1953)). By late July, within a few weeks of bitting,
success from eggs laid 10 August-10 November 1986
intensity had increased to 2068 lice (S.D. = 1970,
was 2-3 (S.D. = 0823), compared to a mean of
N = 86) and overdispersion was considerably re-
30 (S.D. = 1-56) from eggs laid over the same period
duced (Fig. 2B). The maximum intensity of lice on
in 1987 (N = 10 breeding pairs, Wilcoxon T= 8-0,
bitted birds did not exceed natural intensities for
P = 015). Note that the trend was for more young to
birds in feral populations (Clayton, 1989).
be fledged after bitting and fumigation.
Fumigation successfully controlled lice but did
Second, the reproductive success of (unbitted)
not eliminate them completely. The 20 low-load
fumigated birds in the mite experiment did not differ
pairs had a mean intensity of 66 lice (S.D. = 37),
from that of (bitted) fumigated birds in the louse
compared to a mean of 2153 lice (S.D. = 1558) on the
experiment. Unbitted pairs successfully fledged an
23 high-load pairs (U = 0, P = 00001).
average of 0-64 (S.D. = 081) surviving young, com-
Repeated fumigation maintained similar pro-
pared to a mean of 0-75 (S.D. = 091) produced by
portions of high- and low-load birds throughout the
bitted birds (U = 1045, P = 0-80). These results
study (Fig 2C). Of birds remaining in the enclosure
further indicate that bits had no significant effect on
during the follow-up period, the 10 low-load pairs
reproductive success in the captive flock. They also
had a mean intensity of 75 lice (S.D. = 46), compared
serve as additional evidence against seasonal vari-
to a mean of 1573 lice (S.D. = 764) on the 12 high-
ation in reproductive success in spring compared to
load pairs (U = 0, P = 0-0001).
late summer.
The louse intensities of the offspring of low- and
Pyrethrum had no side-effects on the host.
high-load pairs also differed significantly. Young of
Recently hatched young drenched with pyrethrum
fumigated pairs had a mean intensity of 2-2 lice (S.D.
had similar growth rates to those drenched with
= 2-77, N = 5 nests), compared to a mean of 235
water (7 fumigated, 8 sham-fumigated; F = 022,
lice (S.D. = 10-41, N = 4 nests) on young of sham-
P = 064) and all individuals fledged. Post-fledging
fumigated pairs (U = 0, P = 0-01). Once infested,
body mass for the two groups was also similar
the mean number of lice on nestlings correlated with
(F = 0-17, P = 0-69), as were post-fledging survival
the mean number of lice on their parents, regardless
rates (Fisher Exact P = 0-35).
of their age (Kendall T = 0-44, P = 0 0 1 , N = 17
nests; Fig. 3A). The robustness of this relationship
increased after controlling for age (T = 0-5, P = Effects of mites and lice on host fitness
0-005, N = 17 nests; Fig. 3B). The correlation of
Reproductive success of birds at low- and high-load
louse intensity on parents and offspring reflects the
nests did not differ immediately before the mite
vertical transmission of lice documented by Clayton
experiment (12 February-12 March). Prior to
& Tompkins (1994).
treatment low-load pairs produced a mean of 027
(S.D. = 0-47) surviving fledglings, whereas high-load
Effects of season, fumigant and bits on host fitness
pairs produced a mean of 0-91 (S.D. = 1-04) survivors
The reproductive success of parasite-free birds in (U = 42, P = 016). Thus, mite infestation was not
the spring of 1986 did not differ significantly from correlated with host fitness a priori. Note that the
that in late summer. The mean number of young trend was for low-load birds to have poorer re-
fledged from eggs laid April-May 1986 was 2-5 (S.D. productive success than high-load birds prior to
= 1-32), compared to a mean of 2-4 (S.D. = 1-43) treatment.
D. H. Clayton and D. M. Tompkins 200

20 -i A
A
1-6 • •
* •
30-

Log (nestling intensity

i

1-2 •
t-requency

•
o
1

0-8 •

0-4 •

10-
0 -
0-8 1-2 1-6 2-

o -L ILl n n
R , ,
Log (adult intensity + 1)
2000 4000 6000 8000 10000 12000
1 -U - B
Louse intensity
40 -i
•
B 0-6 - • «
•
•
30-
•£ °2• • •
3 •
in •
IE -0-2 - • •
• •
*
-0-6-
•
10 " -10-
0-8 1-2 1-6 20
Log (adult intensity + 1)
fh rffi Fig. 3. (A) Relationship of mean intensity of lice on
nestlings (JV =17 infested nests) to mean intensity of
2000 4000 6000 8000 10000 12000
Louse intensity
lice on their parents; (B) as for (A) but using residuals
40 -, from a plot of louse intensity on nestling age to control
for body size.

using data from nests that were mite-free until

30 -
nestlings reached at least 15 days of age. Seven
infested nests (mean intensity = 3-29, S.D. = 236)
and 7 uninfested nests met this criterion. Analysis of
§ 20 this smaller sample permitted a more direct com-
IT
parison of the impact of mites to the impact of lice,
given that the latter did not infest birds under the age
of 15 days (Clayton & Tompkins, 1994).
Egg laying. Mites had no effect on egg production.
The number laid by the 11 low-load pairs (mean =
1-64, S.D. = 0-81) did not differ significantly from the
2000 4000 6000 8000 10000 12000
number laid by the 11 high-load pairs (mean = 2-0,
Louse intensity S.D. = 0-89; U = 50-5, P = 0-33).
Fig. 2. (A) Frequency distribution of lice in May, prior Lice also had no effect on egg production. During
to bitting (N = 86 birds); (B) distribution of lice in late the main experiment the number of eggs laid by 20
July, several weeks after bitting; (C) distribution of lice low-load pairs (mean = 1-75, S.D. = 0-79) did not
at the end of the study, several months after fumigation. differ significantly from the number laid by 23 high-
( • ) Fumigated low-load birds; ( • ) sham-fumigated load pairs (mean = 161, S.D. = 1-00; U = 2065,
high-load birds. P = 0-45). During the 2-month follow-up period
the number laid by 10 low-load pairs (mean = 2-2,
The impact of mites on post-hatching components S.D. = 1-03) did not differ significantly from the
of fitness was analysed in two ways: (1) using data number laid by 12 high-load pairs (mean = 2-67,
from all 22 nests in the main experiment and (2) S.D. = 1-30; U = 48, P = 0-38).
Effects of ectoparasites on reproduction 201

30n 400-1
D Eggs laid
iH Eggs hatched 350
25
300-

20-
2250
f, 200-1
E 15-
5 150

100-

50
5-
0
3 7 11 15 19 23
Nestling age (days)
Controls Experimental
400-1
14

• Eggs laid
350 I
14
H Eggs hatched 15
40- 300-

25
o> °
30- 200-

150

20- 100-

50-
10-
7 11 15 19 23
Nestling age (days)
Fig. 5. Mean ( + S.E.) nestling growth for low-load (O)
Controls Experimental
and high-load ( # ) nests in (A) mite experiment, where
Fig. 4. Hatching success contingent upon number of N — 9 low-load nests and 3 high-load nests unless
eggs laid in (A) mite experiment and (B) louse otherwise indicated; (B) louse experiment, where
experiment. Controls are low-load nests/birds; N = 15 low-load nests and 16 high-load nests unless
experimental are high-load nests/birds. otherwise indicated.

Hatching success. Mites had a striking effect on when all data were included (F = 0-00, P = 1 0 0 ,
hatching success (Fig. 4A; #2 = 17-38, P= 00001). Fig. 5B), as well as when the analysis was limited to
In contrast, lice had no effect on hatching success data collected prior to Day 23 (F = 1-41, P = 0-24).
(Fig. 4B; x2 = 017, P = 0-91). During the follow-up Lice also had no effect on nestling growth during the
period low-load pairs hatched 17 of 22 eggs laid and follow-up period (F = 0-05, P = 083).
high-load pairs hatched 22 of 32 laid (x2 = 0-47,
P=0-71). Fledging success. Mites had no significant effect on
fledging success (Fig. 6A; Fisher Exact P = 077).
Nestling growth. Mites reduced nestling growth rates. Lice also had no significant effect on fledging success
Although there was no significant effect of mites (Fig. 6B; x2 = 0-53, P = 076). During the follow-up
when all data were included (F = 068, P = 0-41), period 16 of 17 birds fledged from low-load nests and
limiting the analysis to data collected prior to Day 23 18 of 22 birds fledged from high-load nests (Fisher
revealed a significant impact despite extremely small Exact P = 0-26).
sample sizes (F = 3-91, P = 005, Fig. 5A). The
justification for limiting the analysis was that 1 of Post-fledging weight. Mites had no effect on post-
only 3 high-load nests was lost from the sample fledging weights in the main experiment (F = 0,
owing to mortality by Day 23. A similar trend was P = 10) but the analysis had to be truncated after the
apparent for the analysis restricted to birds infested first post-fledging weighing because two-thirds of
after 15 days of age (F = 2-43, P = 0-12). high-load birds died thereafter. Mites did signifi-
In contrast, lice had no effect on nestling growth. cantly reduce post-fledging weights in the analysis
The growth of low-load birds was similar to that of restricted to birds infested after 15 days of age
high-load birds during the main experiment, both (F = 4-06, P = 005).
D. H. Clayton and D. M. Tompkins 202
25-i
D Nestlings fledged
• Eggs hatched H Fledglings surviving
20 HI Nestlings fledged 15-

15-
10-

10-

5-
5-

Controls Experimentals Controls Experimentals

Q Nestlings fledged
D Eggs hatched
i ] Fledglings surviving
40- M Nestlings fledged 30-

30-
20-

20

10-

Controls Experimentals Controls Experimentals

Fig. 6. Fledging success contingent upon number of Fig. 7. Post-fledging survival contingent upon number
eggs hatched in (A) mite experiment and (B) louse of nestlings fledged in (A) mite experiment and (B)
experiment. Controls are low-load nests/birds; louse experiment. Controls are low-load nests/birds;
experimentals are high-load nests/birds. experimentals are high-load nests/birds.

Lice had no significant effect on post-fledging pair (S.D. 0-98; U = 223, P = 0-85). During the
weights during the main experiment (F = 0-00, follow-up period low-load birds produced a mean of
P = 0-96) nor during the follow-up period (F = 0-89, 0-60 survivors (S.D. = 0-84)/pair; high-load birds
P = 0-38). produced a mean of 0-75 survivors/pair (S.D. = 0-96;
U = 56, P=0-76).
Post-fledging survival. Mites reduced post-fledging Over the course of the main experiment, offspring
survival (Fig. 7A; Fisher Exact P = 0-034). Over the survival was significantly correlated with mite in-
course of the experiment, the low-load pairs pro- tensity (Kendall T = - 0 2 1 , P = 002), whereas
duced a mean of 0-64 survivors (S.D. = 0-81)/pair, louse intensity was not correlated with offspring
whereas high-load pairs produced no survivors. For survival in either the main experiment (Kendall
the analysis restricted to young infested after 15 days T = 0003, P = 0-97) or during the follow-up period
of age, 7 of 14 fledglings from uninfested nests ( T = 0 1 2 , P = 0-31).
survived the 6-week period, compared to only 1
of 13 fledglings from infested nests (Fisher Exact Adult weight and survival. Mites had no significant
P = 002). effect on adults. There were no deaths over the
Lice had no effect on post-fledging survival (Fig. course of the experiment and group weights did not
7B; x2 = 0-93, P = 0-50). During the follow-up differ significantly. Low-load birds lost a mean of
period 6 of 16 low-load birds survived compared to 13-32 g (S.D. = 25-25), compared to a mean loss of
9 of 18 high-load birds (x2 = 054, P = 0-70). During 12-55 g (S.D. = 211) by high-load birds (Unpaired,
the main experiment low-load birds produced a one-tailed t = 011, P = 046). Lice also had no effect
mean of 0-75 surviving fledglings/pair (S.D. = 0-91); on adult weight or survival. There were no deaths
high-load birds produced a mean of 0-83 survivors/ during the experiment or during the follow-up
Effects of ectoparasites on reproduction 203

period and lice had no persistent impact on adult was, in turn, influenced by the pattern of infestation.
weight. Although low-load birds tended to gain This did not appear to be the case. Mites first
more weight than high-load birds by the October colonized nest boxes next to a crack in the wall of the
weighing (F = 3-40, P = 0-07), there was no overall barn that was the source of the mites. Over the next
effect of lice on adult weights by the February 1988 few days they spread rapidly to boxes adjacent to
weighing (F = 0-88, P = 035). infested boxes significantly more often than to boxes
Most adults continued to roost in the enclosure adjacent to uninfested boxes (Clayton & Tompkins,
after the removal of window coverings in late January 1994). Thus, the pattern of infestation was dictated
1988. In mid-February all birds still roosting in the by the physical position of the nest boxes, which had
enclosure were captured and weighed. Similar no significant effect on the reproductive success of
proportions of low- and high-load birds were present birds immediately before the experiment (see Results
from the louse experiment (29 of 40 low-load birds section), nor during the previous year (unpublished
and 32 of 46 high-load birds, X2 = OO9, p = 0-95). data).
Lice had no significant effect on the susceptibility Mites were horizontally transmitted among nest
of birds to predation by racoons (Procyon lotor), boxes and so did not rely on contact between parent
which are common predators of feral pigeons in hosts and their offspring for transfer (Clayton &
North America (Johnston, 1992). After removal of Tompkins, 1994). Rapid horizontal transmission
the window coverings, racoons occasionally captured caused an exponential increase in prevalence during
roosting birds and by the end of the study we had the early weeks of the study. Lice, on the other hand,
recovered the banded remains of 9 victims. Four of depended on direct contact between parents and
these were low-load birds and 5 were high-load birds their offspring for transfer, and so were vertically
(f = 0-17, P = 0-96). transmitted. In fact, lice depended on direct contact
between the feathers of parents and their offspring,
as revealed by the timing of transmission in relation
DISCUSSION
to feather emergence (Clayton & Tompkins, 1994).
Tests of the impact of parasites on host fitness are Lice were never observed on the skin of a host. The
often complicated by the fact that parasites are correlation of parent and offspring louse intensity
generally overdispersed among host individuals further shows that vertical transmission was the
(Anderson & Gordon, 1982). In such cases pro- main, if not only, source of infestation. Of course,
hibitively large samples of hosts may be required to heritable resistance to lice could contribute to a
detect significant effects of parasites. One way parent-offspring correlation in louse intensity. How-
around this problem is to convert the overdispersed ever, heritable factors could not have been important
distribution of parasite intensity into a more even in this case since experimentally induced variation in
distribution by manipulating parasite intensities louse intensity was randomly assigned among adult
within the range of natural variation. This approach breeding pairs.
enhances the statistical power of experiments Mites were extremely virulent. Without exception,
designed to test the effects of parasites on host the reproductive success of adults at nests with more
fitness, and circumvents problems of interpretation than just a few mites was zero. The effect of mites
inherent in correlational data. was greatest early in the nesting cycle. Although
In this study louse intensities were manipulated mites had no impact on the number of eggs laid, they
through bitting and fumigation to produce roughly reduced hatching success by more than 7 5 % . The
equal numbers of high- and low-load nests. Mite reason for this effect was that birds at heavily
intensities needed to be manipulated only by infested nests were too agitated to incubate, as shown
fumigation to produce high- and low-load nests. The for Dermanyssus infesting other species of birds
combination of techniques created similar mite and (Moss & Camin, 1970). Adults at heavily infested
louse intensities among the birds in each experiment. nests responded with frequent turning, scratching,
Fumigant had no side-effects on the birds them- preening, leg-shaking and they spent less time
selves. Bits had no apparent side-effects; however, incubating than birds at relatively uninfested nests
every adult bird in the enclosure was bitted to ensure (D. H. Clayton, personal observation). Physical
equal treatment. There was no significant seasonal irritation has been shown to reduce hatching success
variation in reproductive success between the periods in other bird-ectoparasite systems (Duffy, 1983;
in which the mite and louse experiments were Brown & Brown, 1986; Emlen, 1986).
conducted. Mites also reduced nestling growth (Fig. 5A) and
Fumigation to control louse loads was randomized. post-fledging survival (Fig. 7A). These effects may
However, this was not the case for mite loads, which have been the result of anaemia, which is known to
were controlled according to the pattern of mite be caused by Dermanyssus (Kirkwood, 1967) and
infestation (see Materials and Methods section). This other mites (Matthysse, Jones & Purnasiri, 1974;
raises the possibility that the impact of mites on host Clark, 1991). Mites may also have had an indirect
reproduction was influenced by a third variable that effect on the hosts by vectoring endoparasites. D.
D. H. Clayton and D. M. Tompkins 204

gallinae has been shown to vector a variety of data on the impact of mites under field conditions.
protozoan, bacterial and viral diseases of birds However, the negative effects of other species of
(Moss, 1978). We did not investigate the proximal Dermanyssus on wild birds (Moss & Camin, 1970)
cause(s) of the impact of mites on growth and suggest that our findings are not an artifact of captive
survival. conditions.
Lice had no effect on host fitness, despite a The dissimilar effects of mites and lice could not
prolonged period of data collection with larger have been due to seasonal differences as the re-
sample sizes than in the mite study. Lice had no productive success of control birds in the two
impact on the number of eggs laid or on hatching experiments was quite similar. The dissimilar effects
success. This result is not surprising, given that of mites and lice were also unlikely to have been
Ischnocera on rock doves do not cause irritation. artifacts of different parasite population sizes. Sev-
Birds with large numbers of lice do not preeen or eral pairs of adult birds with thousands of lice
scratch more than those with low numbers because produced the maximum of two surviving offspring
the lice in question never go on the skin of the host during the main louse experiment. In contrast, only
(Clayton, 1990). In the current study birds with high nests with extremely low mite loads produced any
louse intensities appeared to incubate as consistently surviving offspring.
as those with low intensities, although we did not The relative benevolence of Ischnocera appears to
quantify incubation times. have at least two main components. First, restriction
Lice also had no effect on nestling growth, fledging to feathers reduces irritation to the skin of the host,
success, post-fledging survival, nor the weight or leading to more consistent incubation of eggs.
survival of adults. These results are interesting, Ischnocera have evolved a specialized niche re-
given that louse-induced feather damage is known to stricted to feathers, derived from a more generalized
have energetic costs even at relatively warm ambient niche on the body of the host (Waage, 1979). The
temperatures. Booth et al. (1993) documented an latter was, in turn, probably derived from an even
8-5 % increase in the metabolic rates of bitted, sham- more generalized niche in the host's nest (Waage,
fumigated rock doves over a range of ambient 1979). Specialization to feathers carries benefits such
temperatures, including 22 °C - the mean tempera- as a predictable food supply. But it also carries costs
ture in our study. The amount of feather damage such as limited ability to disperse to new hosts. Our
caused by lice in our study (reported by Clayton, data, in conjunction with a survey of the literature
1990) was comparable to that in the study of Booth (Clayton & Tompkins, 1994), show that limited
et al. (1993). Although the latter was primarily a field transmission is linked to reduced virulence, as
experiment, birds were taken captive and fed ad suggested by theoretical work (Anderson & May,
libitum for several days prior to the measurement of 1982; Ewald, 1983; Lehmann, 1993).
metabolic rates in the laboratory. Birds in our study Second, dependence on host reproduction for
were fed ad libitum throughout, which may have transmission means that feather specialists, such as
allowed them to compensate for any energetic cost of Ischnoceran lice or feather mites (Ateyo & Gaud,
feather damage. 1979), are unlikely to serve as vectors of virulent
Such compensation is one explanation for why endoparasites (which could occur via ingestion of
there was no significant effect of lice on winter infected lice during preening, analogous to the
survival in our study. High-load birds in the study ingestion of cestodes by dogs during grooming
of Booth et al. (1993) had significantly lower body (Grundy, 1981)). Until recently (Bartlett, 1993),
mass than low-load birds, linked to a significant Ischnoceran lice had not been recorded as endo-
reduction in the over-winter survival of high-load parasite vectors (Clayton, 1990). Pigeon Ischnocera
birds in earlier work (Clayton, 1989). The pigeons in have not been recorded as vectors, despite the fact
our study had ready access to shelter and our study that Columbicola columbae is a relatively well studied
was not as long term as the earlier one (Clayton, 'white rat' (Eichler et al. 1972). Transmission may
1989). thus constrain vector biology just as it appears to
Although conditions in the enclosure were more constrain more direct components of virulence. For
favourable than those in the field, our test of the sake of comparison, Amblyceran lice, which are
impact of lice on host reproductive success appears more mobile than Ischnocera, are known to vector
to be an accurate reflection of the natural situation. A several kinds of endoparasites (Clayton, 1990).
field experiment modelled on the one in this paper The results of our mite experiment support the
revealed no impact of lice on the reproductive claim that ectoparasites can be costly, as emphasized
success of free-ranging birds (Clayton, 1989). Hence, by Lehmann in an important review (1993). The
results for the captive flock are not merely a by- widespread assumption that ectoparasites are benign
product of more favourable conditions in the en- is clearly an oversimplification. However, the results
closure. The severe impact of mites in this study of our louse experiment indicate that some ecto-
further indicates that favourable conditions do not parasites are, in fact, benign with regard to host
preclude the detection of parasite effects. We have no reproduction. It is therefore dangerous to draw
Effects of ectoparasites on reproduction 205

generalizations about costs of ectoparasitism, just as CLAYTON, D. H. (1990). Mate choice in experimentally
it would be dangerous to draw generalizations about parasitized rock doves: lousy males lose. American
costs of endoparasitism. Zoologist 30, 251-62.
CLAYTON, D. H. (1991). Coevolution of avian grooming
and ectoparasite avoidance. In Bird—Parasite
Interactions: Ecology, Evolution, and Behaviour (ed.
Funds were provided by N S F Grant BSR-8612575, N I H
Grant GM07197, Kaytee Products Inc., and L. Getz. Loye, J. E. & Zuk, M.), pp. 258-289. Oxford: Oxford
D . M . T . was supported by a NERC studentship. Data University Press.
were collected by D . H . C . as part of doctoral work at the CLAYTON, D. H., GREGORY, R. D. & PRICE, R. D. ( 1 9 9 2 ) .
University of Chicago. We are grateful to N. Burley, C. Comparative ecology of Neotropical bird lice (Insecta:
Cooper, L. Getz, P. Mankin, and D. Wake for facilities or Phthiraptera). Journal of Animal Ecology 61, 781-95.
accommodations. We are particularly grateful to D. Droge CLAYTON, D. H. & TOMPKINS, D. M. (1994). Ectoparasite
for starting the pigeon colony and allowing us to work with
it. We thank J. Howe and Jean and Jim Nicholson for virulence is linked to mode of transmission.
technical assistance, as well as S. Arnold, T . Ateyo, R. Proceedings of the Royal Society of London, B 256,
Clayton, R. Clayton Jr., J. Fitzpatrick, J. Kethley, R. 211-17.
Lande, M. Lloyd, A. & T . Peterson, M. Phelan, D. DERYLO, A. (1974a). Studies on the economic
Schemske, J. West, and J. & L. White for other assistance harmfulness of Mallophaga. I. T h e influence of
or discussion. Thanks also to N. Gotelli, P. Harvey, A. invasion of lice on the state of health in hens and
Herre and A. Read for comments on the manuscript. turkeys (in Polish). Medycyna Weterynaryjna 30,
D. H. C. especially thanks K. Clayton for assisting with
353-7.
all phases of the study and for tolerating numerous con-
versations in public about 'my lice'. DERYLO, A. (19746). Studies on the economic
harmfulness of Mallophaga. II. Influence of lice
infestation on egg laying and hatching of hens (in
Polish). Medycyna Weterynaryjna 30, 406-10.
REFERENCES
DEVANEY, j . A. (1976). Effects of the chicken body louse,
ANDERSON, R. M. & GORDON, D. M. (1982). Processes Menacanthus stramineus, on caged layers. Poultry
influencing the distribution of parasite numbers Science 55, 430-5.
within host populations with special emphasis on DROGE, D. L. (1986). Parent-offspring conflict and the
parasite-induced host mortalities. Parasitology 85, allocation of parental investment in the feral pigeon,
373-98. Columba livia. Ph.D. thesis, University of Illinois,
ANDERSON, R. M. & MAY, R. M. (1982). Coevolution of Urbana-Champaign.
hosts and parasites. Parasitology 85, 411-26. DUFFY, D. c. (1983). The ecology of tick parasitism on
ATEYO, w. T. & GAUD, j . (1979). Feather mites and their densely nesting Peruvian seabirds. Ecology 64, 110—19.
hosts. In Recent Advances in Acarology (ed. EICHLER, W . , ZLOTORCZYKA, J., LUDWIG, H. W. & STENRAM,
Rodriquez, J. G.), pp. 355-361. New York: Academic H. i. (1972). The pigeon louse Columbicola columbae
Press.
columbae (German translation). Angewandte
BARTLETT, c. M. (1993). Lice (Amblycera and
Parasitologie 13, 1-18.
Ischnocera) as vectors of Eulimdana spp. (Nematoda:
EMLEN, J. T. (1986). Responses of breeding cliff swallows
Filarioidea) in Charadriiform birds and the necessity
to nidicolous parasite infestations. The Condor 88
of short reproductive periods in adult worms. Journal
110-11.
of Parasitology 79, 85-91.
EWALD, p. w. (1983). Host-parasite relations, vectors,
BENBROOK, E. A. (1965). External parasites of poultry. In
and the evolution of disease severity. Annual Review
Diseases of Poultry (ed. Biester, H. E. & Schwarte,
L. H.), pp. 925-964. Iowa: Iowa State University of Ecology and Systematics 14, 465—85.
Press. GRUNDY, j . H. (1981). Arthropods of Medical Importance.
BLISS, c. i. (1953). Fitting the negative binomial Chilbolton, Hampshire: Noble Books Ltd.
distribution to biological data. Biometrics 9, 176—200. HURLBERT, s. H. (1984). Pseudoreplication and the design
BOOTH, D. T., CLAYTON, D. H. & BLOCK, B. A. (1993). of ecological field experiments. Ecological Monographs
Experimental demonstration of the energetic cost of 54, 187-211.
parasitism in free-ranging hosts. Proceedings of the JACKSON, j . A. (1985). On the control of parasites in nest
Royal Society of London, B 253, 125-9. boxes and the use of pesticides near birds. Sialia 7,
BROWN, c. R. & BROWN, M. B. (1986). Ectoparasites as a 17-25.
cost of coloniality in cliff swallows (Hirundo JOHNSTON, R. F. (1992). Rock dove. In The Birds of
pyrrhonota). Ecology 67, 1206-18. North America (ed. Poole, A., Stettenheim, P. & Gill,
CLARK, L. (1991). The nest protection hypothesis: the F.). Washington: The American Ornithologists'
adaptive use of plant secondary compounds by Union.
European starlings. In Bird-Parasite Interactions : KIRKWOOD, A. C. (1967). Anaemia in poultry infested
Ecology, Evolution, and Behaviour (ed. Loye, J. E. & with the Red Mite Dermanyssus gallinae. Veterinary
Zuk, M.), pp. 205-221. Oxford: Oxford University Record 80, 514-15.
Press. LEHMANN, T. (1991). Ectoparasitic impacts on Gerbillus
CLAYTON, D. H. (1989). Coevolutionary ecology of the andersoni allenbyi under natural conditions.
rock dove, Columba livia, and its chewing lice Parasitology 104, 479-88.
(Mallophaga: Ischnocera). Ph.D. thesis, University of LEHMANN, T. (1993). Ectoparasites: direct impact on host
Chicago. fitness. Parasitology Today 9, 8-13.
D. H. Clayton and D. M. Tompkins 206

LEVI, w. M. (1957). The Pigeon. Columbia, S.C.: R. L. MOSS, w. w. & CAMIN, j . H. (1970). Nest parasitism,
Bryan Co. productivity, and clutch size in Purple Martins.
MARGOLIS, L., ESCH, G. W., HOLMES, J. C , KURIS, K. M. & Science 168, 1000-3.
SCHAD, G. A. (1982). The use of ecological terms in NELSON, B. c. & MURRAY, M. D. (1971). The distribution
parasitology. Journal of Parasitology 68, 131-3. of Mallophaga on the domestic pigeon (Columba
MARSHALL, A. G. (1981). The Ecology of Ectoparasitic livid). International Journal for Parasitology 1, 21-9.
Insects. London: Academic Press. SIKES, R. K. & CHAMBERLAIN, R. w. (1954). Laboratory
MATTHYSSE, G., JONES, C. J. & PURNASIRI, A. (1974). observations on three species of bird mites. Journal of
Development of northern fowl mite populations on Parasitology 40, 691-7.
chickens, effects on the host, and immunology. Search STRANDTMANN, R. W. & WHARTON, G. W. (1958). A
Agriculture 4. Manual of Mesostigmatid Mites Parasitic on
MOSS, w. w. (1978). The mite genus Dermanyssus: a Vertebrates. Institute of Acarology, University of
survey, with description of Dermanyssus trochilinis, Maryland.
n.sp., and a revised key to the species (Acari: WAAGE, j . K. (1979). The evolution of insect/vertebrate
Mesostigmata: Dermanyssidae). Journal of Medical associations. Biological Journal of the Linnean Society
Entomology 14, 627-40. 12, 187-224.