Seasonal patterns in Tree Swallow prey (Diptera) abundance are affected by agricultural intensification

Ecological Applications, 23(1), 2013, pp. 122–133� 2013 by the Ecological Society of America

Seasonal patterns in Tree Swallow prey (Diptera) abundance areaffected by agricultural intensification

SEBASTIEN RIOUX PAQUETTE,1 DANY GARANT, FANIE PELLETIER, AND MARC BELISLE

Departement de Biologie, Universite de Sherbrooke, Sherbrooke, Quebec, J1K 2R1 Canada

Abstract. In many parts of the world, farmland bird species are declining at faster ratesthan other birds. For aerial insectivores, this decline has been related to a parallel reduction inthe abundance of their invertebrate prey in agricultural landscapes. While the effects ofagricultural intensification (AI) on arthropod communities at the landscape level have beensubstantially studied in recent years, seasonal variation in these impacts has not beeninvestigated. To assess the contention that intensive cultures negatively impact food resourcesfor aerial insectivorous birds, we analyzed the spatiotemporal distribution patterns of Diptera,the main food resource for breeding Tree Swallows (Tachycineta bicolor), across a gradient ofAI in southeastern Quebec, Canada. Linear mixed models computed from a data set of 5000samples comprising .150 000 dipterans collected over three years (2006–2008) suggest thatboth Diptera abundance and biomass varied greatly during swallow breeding season,following a quadratic curve. Globally, AI had a negative effect on Diptera abundance (but notbiomass), but year-by-year analyses showed that in one of three years (2008), dipterans weremore abundant in agro-intensive landscapes. Analyses also revealed a significant interactionbetween the moment in the season and AI: In early June, Diptera abundances were similarregardless of the landscape, but differences increased as the season progressed, with highlyintensive landscapes harboring fewer prey, possibly creating an ‘‘ecological trap’’ for aerialinsectivores. While global trends in our results are in agreement with expectations (negativeimpact of AI on insect abundance), strong discrepancies in 2008 highlight the difficulty ofpredicting the abundance of insect communities. Our study indicates that predicting the effectsof AI may prove more challenging than generally assumed, even when large data sets arecollected, and that temporal variation within a season is important to take into consideration.While further work is required to assess the direct impacts of these seasonal trends in Dipteraabundance on bird breeding success and post-fledging survival, management strategies inagricultural landscapes may need to consider the phenology of breeding birds and theirdipteran prey in order to mitigate the potentially negative effects of AI late in the breedingseason.

Key words: agro-intensive landscapes; biomass; Diptera abundance; linear mixed modeling; southeast-ern Canada; spatiotemporal analyses; Tachycineta bicolor; Tree Swallow.

INTRODUCTION

Over the last several decades, the exponentially

increasing human population has put tremendous

pressure on agriculture, resulting in the rapid evolution

of this industry towards greater efficiency (Blaxter and

Robertson 1995). This trend, often referred to as

agricultural intensification (AI), is associated with key

changes in agricultural practices (including improved

land drainage, hedgerow removal, the rise of agrochem-

ical inputs, and earlier planting and harvesting; Donald

et al. 2001, Robinson and Sutherland 2002), which have

transformed farmlands into structurally simplified eco-

systems harboring reduced biodiversity (e.g., Benton et

al. 2003, Tscharntke et al. 2005, Batary et al. 2010).

However, temporal variation in the effects of AI within

seasons has not been investigated, even though success-

ful management schemes in agricultural landscapes

often depend on adequate knowledge of the phenology

of species (Perlut et al. 2006). Here, we assess the

spatiotemporal effects of AI on the abundance of insects

from the order Diptera, the main food resource of

breeding Tree Swallows, during the nesting and fledging

stages of this insectivorous bird.

Global declines in the abundance of bird species in

most parts of the world are attributed to diverse factors

including habitat loss (Robinson et al. 1995), increasing

predation (Bohning-Gaese et al. 1993), or even extreme

weather events (Sauer et al. 1996). Interestingly,

farmland birds are among the most steeply declining

birds (Donald et al. 2001, Murphy 2003), especially

aerial insectivores, both in Europe and North America

(e.g., Benton et al. 2002, McCracken 2008). Species in

this guild exhibit a wide range of ecological and life

history traits, so that the most likely hypothesis to

Manuscript received 12 January 2012; revised 29 June 2012;accepted 3 July 2012. Corresponding Editor: R. L. Knight.

1 E-mail: [email protected]

122

explain their common decline is a large-scale decline in

their most important food resource, aerial insect

populations (Nebel et al. 2010). Indeed, there is a

growing body of evidence documenting the negative

effects of AI on the abundance and diversity of

invertebrates (e.g., Aebischer 1991, Elliott et al. 1998,

Vickery et al. 2001, Conrad et al. 2006, Attwood et al.

2008, O’Rourke et al. 2011), and this worrisome trend is

believed to be responsible for the important decline

observed in many farmland bird species (Krebs et al.

1999, Donald et al. 2006). AI is hypothesized to act

indirectly on bird population abundance by reducing the

abundance of their prey, either through pesticide use or

habitat homogenization (e.g., Benton et al. 2002, Evans

et al. 2007, Shortall et al. 2009). For instance, Evans et

al. (2007) found that aerial insect abundances were more

than three times greater over pastures than over cereal

fields in southern Britain. Yet, many studies of insect

prey abundance in agricultural landscapes typically

focus on habitat impacts without accounting for

possible seasonal variations in habitat/insect community

relationships (e.g., Wickramasinghe et al. 2004, Bengts-

son et al. 2005, Fuentes-Montemayor et al. 2011).

Because the spreading of chemicals in intensively

managed fields is not performed evenly throughout the

season and is linked to both crop and pest phenology,

the effects of intensive agriculture on insect communities

may vary in time. Such temporal variation could

represent an important determinant of the ecology of

insects in cultivated landscapes and a challenge for aerial

insectivores nesting in agricultural habitats.

In this article, we analyze the spatiotemporal distri-

bution patterns of Diptera among a network of 40 farms

located along a gradient of AI in Quebec, Canada. Our

main objective was to describe abundance patterns of

aerial insect prey in this region, not only spatially, but

also temporally, relying on an exhaustive sampling

scheme performed continuously during June and July

of three different years (2006, 2007, and 2008). This

study system was selected because the breeding success

of Tree Swallows Tachycineta bicolor (Vieillot 1808)

nesting on these farms has been monitored since 2004 on

the same spatial scale. The Tree Swallow is no exception

among aerial insectivores, as populations have declined,

on average, by 2.5% per year in Canada between 1986

and 2006 (McCracken 2008). At the same time, the

transition from traditional cultures (extensive cultures

like hayfields, pastures, and fallows) to intensive cultures

(mostly maize, soybeans, and other cereals) has been

very profound. For instance, in 2006, maize cultures in

Quebec occupied 14 times the territory they occupied 50

years earlier (total 485 000 ha; Statistics Canada 2006).

Previous analyses of the data collected in our system

have revealed the multifaceted negative impacts of

intensive agriculture on breeding success in Tree

Swallows. Ghilain and Belisle (2008) have shown that

clutch size is only weakly affected by the level of AI, yet

individuals breeding in areas with a high proportion of

extensive cultures (45%) fledge twice more young than

those breeding in areas mainly composed of intensivecultures (10% extensive cultures). These results support

the hypothesis that agro-intensive landscapes harborlower levels of invertebrate prey, in turn affecting the

survival of chicks before fledging. Thus, our study site isideal to investigate changes in prey abundance, as the

relationship between AI and prey abundance remains tobe assessed. More specifically, here we: (1) analyzedtemporal variation in prey abundance and biomass

during key phases of the swallow breeding season (withJune and July corresponding to nestling, fledging, and

post-fledging periods), and (2) quantified the influence oflandscape structure (proportion of intensive cultures) on

the temporal patterns of prey abundance and biomassacross the intensification gradient. Our main expectation

was that the effects of intensification on aerial insectswould increase through the season, as the use of

chemical inputs in intensive cultures likely increases. Asurvey of spraying practices in Canada revealed that the

majority of farmers using pesticides are spraying thembased on factors such as crop growth stages, first signs

of pests, regional monitoring of pests, or when insectsexceed economic injury levels (Agriculture and Agri-

Food Canada 1998). Thus, one would expect differencesamong intensive and extensive landscapes to increase as

crops grow and pest species emerge throughout theseason.

MATERIALS AND METHODS

Study system

The study area, in southeastern Quebec, Canada, is

characterized by a longitudinal gradient of AI (Fig. 1),where dairy and small-scale farms in the east are

gradually replaced by large-scale continuous rowcropping in the west with increasing input of pesticides

and fertilizers (Jobin et al. 2005). Other noteworthychanges associated with this intensification gradient

include smaller and increasingly fragmented forestcover, accrued drainage of wetlands, and canalization

of streams (Belanger and Grenier 2002). The 40 farmssurveyed in this study delimit an area of ;10 000 km2.

Additional details on the study system are available inGhilain and Belisle (2008) and Porlier et al. (2009).

Aerial insect sampling

Between 2006 and 2008, 80 insect traps were set up on

the 40 selected farms (two traps/farm). They were placedat about one-third and two-thirds of a 450-m linear

transect delineated by 10 Tree Swallow nest boxes so tomaximize the chances that our sampling was represen-

tative of food resources available to swallows. We usedcombined window/water-pan flight traps consisting of a

yellow bucket placed 1.5 m above ground and mountedby two plexiglass screens at right angles to minimize the

influence of wind direction (Duelli et al. 1999). Bucketswere half-filled with salt-saturated water, and detergent

was added to reduce surface tension. While these passive

January 2013 123DIPTERA ABUNDANCE IN AGROECOSYSTEMS

traps might not reflect the total composition of aerialinsect communities in the study area, their capture ratesdo provide a good proxy for aerial prey available to TreeSwallows, as they reflect both the density and the level of

activity of insects (Hoye and Forchhammer 2008).Samples were collected every two days from the first

day of June to the end of July in 2006 and 2007, and to

mid-July in 2008 (see Table 1 for exact dates). Thisperiod overlaps with the nesting, fledging, and post-fledging stages of Tree Swallows (see Appendix A for

data on breeding phenology of swallows in our studysystem). Some samples were lost or discarded due tohandling errors, poor preservation of specimens, or field

conditions (e.g., overflowing of buckets due to heavyrainfall, especially in 2008), representing ;8% of allcollected samples. Captured insects were sorted to the

order. Afterwards, insects from each order were dried at608C for 24 h, and dry biomass was weighed to thenearest 0.0001 g (0.1 mg). We restricted our analyses toDiptera data, as they represent the majority of insect

prey in the diet of Tree Swallow nestlings (Quinney andAnkney 1985, McCarty and Winkler 1999a). Indeed, theanalysis of 108 boluses collected on our study site from

Tree Swallow nestlings during the insect sampling periodrevealed that Diptera represent 62.5% of insects fed tonestlings in terms of abundance and 67.3% of the total

biomass (see Appendix B). Tree Swallows are also

known to exhibit a preference for Odonata (McCarty

and Winkler 1999a), but the proportion of Odonata in

the diet of nestlings is tightly linked to the percentage of

open water within 400 m of the nest (Mengelkoch et al.

2004). Bolus analyses confirmed the scarcity of Odonata

in the diet of nestlings in our study area (0.3% in number

and 1.7% in biomass), which is not surprising given the

rarity of water bodies in the system (the percentage of

water within 500 m of nest boxes was 0.6% 6 1.2%

[mean 6 SD]).

Weather and landscape variables

Temperature data were collected with Thermochron

iButton devices (model DS1922L; Embedded Data

Systems, Lawrenceburg, Kentucky, USA) placed on each

of the 40 farms and set to record temperature every hour

during the sampling period. Precipitation data (rainfall

over previous two days) were collected on each farm with

a pluviometer. Wind speed data were retrieved from two

meteorological stations (Sherbrooke Airport and Saint-

Hubert Airport; available online)2 located at both ends of

our sampling area (close to farms 13 and 33; see Fig. 1).

FIG. 1. Distribution of the 40 farms in southeastern Quebec, Canada, where Tree Swallow (Tachycineta bicolor) nest boxeshave been monitored since 2004 and where insect traps were placed for this study between 2006 and 2008. The proportion ofintensive cultures (light-gray-shaded area) increases from east to west. Numbered circles indicate farm locations.

2 http://climate.weatheroffice.gc.ca/advanceSearch/searchHistoricData_e.html?Prov¼QC&StationID¼9999&Year¼2012&Month¼8&Day¼29&timeframe¼1

SEBASTIEN RIOUX PAQUETTE ET AL.124 Ecological ApplicationsVol. 23, No. 1

Wind data from the closest station to each farm was used

(hourly wind speed from the two stations were highly

correlated, r ¼ 0.75). In order to evaluate the effect of

different agricultural practices on the abundance of

Diptera, we delimited areas corresponding to water

surfaces, tree cover (e.g., forest and hedgerows), extensive

cultures (mostly pastures, hayfields, fallows, and organic

farming), and intensive cultures (all monocultures,

including soybean, maize, and other cereals) using

orthophotos (scale 1:40 000) of each farm within ArcGIS

(ESRI 2008). The relative proportion occupied by each of

these four habitat categories was computed within a

radius of 50 m and 500 m around each insect trap.

Landscape composition was validated every year by

direct observations at each trap site.

Data analyses

Linear mixed models (LMMs) were used to analyze

Diptera abundance and biomass patterns among the 40

farms. Since count data are typically characterized by a

Poisson distribution, we first attempted to analyze

abundance data with generalized linear mixed models

(GLMMs) using a Poisson error distribution and a log

link function. However, most GLMMs failed to

converge, so we decided to perform all analyzes using

LMMs on log-transformed abundance data. This is

mostly inadequate when data sets comprise many ‘‘zero’’

observations (Sileshi et al. 2009), but in our abundance

data, there were no such observations in 2006 and 2007,

and only a few in 2008 (19), which were recorded as an

abundance of 1 and biomass of 0.0001 g before log-

transformation. Conformity of models to assumptions

of independence, homoscedasticity, and normality of

residuals was assessed through visual inspection of

residuals (plots of residuals vs. explanatory variables,

residuals vs. fitted values, and frequency distribution of

residuals) and did not reveal any violation, including the

absence of temporal autocorrelation patterns. Biomass

data were also log-transformed before analysis.

Models were run separately for each of the two

response variables. In each case, a model was initially

computed using data from the three years (hereafter

referred to as a ‘‘global’’ model), with all explanatory

variables: year as a factor (YEAR), day of collection (DAY),

mean temperature in the 48 hours preceding collection

(TEMP), mean wind speed (WIND), and total precipitation

in the 48 hours preceding collection (RAIN). Quadratic

terms of DAY and TEMP were also included. RAIN2 and

WIND2 were excluded from final models because they

never had a significant effect in preliminary analyses.

Interactions between YEAR and DAY, and YEAR and DAY2

were also included. In addition, the global model included

two landscape variables as predictors (the proportion of

intensive cultures within a 50-m radius [INT50] of the trap,

and the same proportion within 500 m [INT500]), as well

as their interaction (INT503 INT500) and their interactions

with DAY, DAY2, and YEAR to account for the possibility

that AI may have different impacts within and across

years. INT50 was included to account for microhabitat

effects on Diptera catches, while INT500 was the focal

landscape predictor of interest. Weather variables were

included to control for their effects across time and space.

The only pair of highly correlated predictors (r . 0.60)

was INT50 and INT500. In all models, farm identity (FARM)

was included as a random effect, allowing random

intercepts and random slopes for both DAY and DAY2 on

each farm. Six additional models comprising subsets of

landscape predictors (INT50, INT500, their interaction, and

their interactions with DAY and DAY2) were computed to

obtain a set of competing models.

Models were computed in R version 2.13.1 (R

Development Core Team 2011) using the lmer function

implemented in the lme4 package (Bates and Maechler

2010). All explanatory variables were centered and

standardized following the approach of Gelman (2008),

as implemented in the arm R package (Gelman et al.

2011), in order to facilitate the interpretation of model

coefficients (Schielzeth 2010). An information theoretic

approach was employed to examine the model set for each

response variable by calculating a sample size-corrected

form of Akaike’s information criterion (AICc) following

recommendations in Burnham and Anderson (2002). For

each set, weighted support for each model was calculated

on the basis of DAICc with the AICcmodavg R package

(Mazerolle 2011). Model averaging was performed with

the MuMIn R library (Barton 2011) using the ‘‘natural

average’’ method, where parameter estimates for each

predictor are averaged only over models which include

this predictor (Burnham and Anderson 2002).

RESULTS

Raw data

In total, we obtained Diptera abundance data from

4970 samples and biomass data from 4906 samples.

Details about sample sizes for each year are provided in

Table 1. Diptera catches on each farm varied substan-

TABLE 1. Sampling effort of Diptera catches performed on 40 farms (two traps/farm) located in southeastern Quebec, Canada,2006–2008.

Year N, abundance N, biomass Total no. individuals No. sampling days First day of sampling� Last day of sampling�

2006 1969 1928 66 107 66 152 (1 Jun) 219 (7 Aug)2007 2066 2048 48 816 56 149 (29 May) 210 (29 Jul)2008 935 930 42 442 46 151 (30 May) 197 (15 Jul)

Notes: Shown are the number of Diptera samples (N ) included for abundance and biomass analyses.� Noted as number of days since 1 January (Julian date), with actual date in parentheses.

January 2013 125DIPTERA ABUNDANCE IN AGROECOSYSTEMS

tially, as revealed by large standard deviations on mean

abundance and biomass on each farm (Fig. 2a, b). There

was no significant correlation between mean Diptera

abundance or biomass and the proportion of intensive

cultures on a farm (Fig. 2a, b). The correlation between

Diptera abundance and biomass in each sample was

highly significant (r¼ 0.47, P , 0.001), but considerable

variation was nonetheless observed (Fig. 2c), stressing

the relevance of performing analyses on both variables

separately. The comparison of early- (samples collected

before Julian day 161; where Julian day is the number of

days since 1 January) and late- (after Julian day 189)

season data, revealed that highly intensive farms

harbored lower Diptera abundance/biomass late in the

season (Fig. 2d). These raw data confirmed the need to

compute LMMs that take into consideration not only

intensive agriculture, but temporal variation as well.

Global models and interannual variation

Models computed on the total data set (three years

combined) revealed significant differences in Diptera

abundance and biomass among the three years, and all

pairwise Tukey’s honest significant difference tests were

significant with P , 0.001, except biomass for the 2006–

2007 comparison (P¼ 0.15). These tests only considered

main effects, but significant interactions between YEAR

and DAY, and between YEAR and INT500, were also

observed. The contributions of competing models to the

final (averaged) models, as well as parameter estimates

from the final models, are presented in Appendix C and

D, respectively. Overall, both Diptera abundance and

biomass followed a quadratic curve during the sampling

season peaking around Julian day 185 or 4 July (Fig.

3a, f ). Abundance was negatively affected by the

proportion of intensive cultures within 500 m (Fig. 3c),

whereas the single effect of INT500 on biomass was not

significant, although it did affect biomass through a

significant interaction with DAY2 (Appendix D). Tem-

perature consistently had a negative effect on Diptera

abundance and biomass (Fig. 3b, g), wind only affected

abundance (Fig. 3e), but precipitation was not a

significant predictor. To simplify interpretation, we

decided to compute separate model sets for each year,

thus calculating six model sets (3 years 3 2 response

FIG. 2. Raw data collected in the present study. Mean Diptera (a) abundance and (b) biomass per trap sample are plotted withrespect to the proportion of intensive cultures within 500 m (INT500) of each trap site (error bars indicate 6SE). Panel (c) illustratesthe relationship between Diptera abundance and biomass within trap samples. In panel (d), mean Diptera abundance and biomassper sample are compared between landscapes with INT500 . 0.6 (solid circles) and those with INT500 , 0.2 (open circles) at thebeginning (Julian day � 160; where Julian day is the number of days since 1 January) and end (Julian day � 190) of the samplingseason. Error bars in this panel indicate 95% confidence intervals (CIs). Results from pairwise Tukey’s honestly significantdifference tests are indicated for landscape comparisons early and late in the season.

SEBASTIEN RIOUX PAQUETTE ET AL.126 Ecological ApplicationsVol. 23, No. 1

variables) following the same methodology. Summary

statistics of all models in each set prior to model

averaging are listed in Appendix E. For all models, there

was a relatively strong positive correlation between

fitted values derived from model coefficients and

empirical values, and the proportion of variation

explained by the models (pseudo R2; Zuur et al. 2009)

varied between 0.24 and 0.43 (Appendix F).

Spatiotemporal patterns in 2006 and 2007

The standardized coefficients of all predictors in final

(averaged) models, as well as their unconditional

standard error and confidence intervals, are presented

in Appendix G (abundance) and Appendix H (biomass).

Abundance and biomass patterns for the first two years

of the study followed the general trends outlined by the

global models: Both DAY and DAY2 were significant

predictors in all models, whereas AI negatively impacted

abundance but not biomass. In addition, significant

interactions were observed between the Julian day and

the amount of intensive agriculture. In order to visualize

the effect of these interactions, we calculated predicted

seasonal curves of Diptera abundance and biomass for

different values of INT500 (ranging from 0 to 1), while

fixing other predictors to their mean (Fig. 4a–d). On this

figure, both abundance and biomass were similar

regardless of the proportion of intensive cultures at the

beginning of the sampling season, namely around Julian

day 150. However, catches of Diptera gradually

diverged among traps, with lower catches in more

agro-intensive landscapes. For instance, at Julian day

190, completely extensive landscapes were expected to

FIG. 3. Significant main effects in Diptera (a–e) abundance and (f, g) biomass global models (computed from the combineddata of 2006, 2007, 2008 [log-transformed]) while fixing other predictors in the models to their mean (using 2007 as reference year).Dashed lines indicate 95% CIs. Julian day is the number of days since 1 January.

January 2013 127DIPTERA ABUNDANCE IN AGROECOSYSTEMS

harbor twice as many individuals as exclusively intensive

landscapes in 2006 (38.5 individuals vs. 19.7 after

conversion of log-transformed data), and 70% more in

2007 (24.5 individuals vs. 14.4).

Spatiotemporal patterns in 2008

Models for 2008 were considerably different than

those for 2006 and 2007, with fewer significant effects

among predictors and larger standard errors, and a

positive coefficient for INT500, indicating that Diptera

abundance and biomass were actually higher in intensive

landscapes in 2008. Significant interactions between

Julian day and AI were also observed in 2008; the

graphical representation of this interaction (Fig. 4e, f )

showed that abundance and biomass in very intensive

landscapes increased in the first half of the sampling

period before decreasing in the second half. On the other

hand, predicted values from landscapes with moderate

or low levels of agro-intensive cultures (,0.5) followed

opposite trends, either decreasing throughout the entire

FIG. 4. Model-averaged predictions of Diptera (a, c, e) log-transformed abundance and (b, d, f ) biomass (originally measuredin grams) throughout the sampling period in (a, b) 2006, (c, d) 2007, and (e, f ) 2008 for different proportions of intensive cultureswithin 500 m (INT500) of each trap site when values of other predictors were fixed to their mean. Julian day is the number of dayssince 1 January.

SEBASTIEN RIOUX PAQUETTE ET AL.128 Ecological ApplicationsVol. 23, No. 1

season or reaching a minimum before slightly increasing

during the second half of sampling period.

DISCUSSION

To our knowledge, our study is one of the first to

explicitly address how insect prey abundance in agricul-

tural landscapes varies both in space (along a gradient of

intensification) and time during the season, with

extensive spatial and temporal sampling. Douglas et al.

(2010) sampled several arthropod orders (including

Diptera) once a month from April to August, but their

study was restricted to barley crops. Gruebler et al.

(2008) did survey aerial insect abundance in different

agricultural habitats, but focused on short-term weather

effects and did not investigate nor detect seasonal

trends. Analyses on our global data set revealed that,

overall, Diptera abundance and biomass followed a

quadratic curve during the months of June and July

(similar to results in Douglas et al. 2010), peaking at the

beginning of July. Global results also showed that the

proportion of intensive cultures within 500 m of insect

traps negatively influenced the abundance of Diptera,

but not total biomass. Year-specific models confirmed

these trends in two of the three years of the study (2006

and 2007). Furthermore, our results emphasized the

importance of the interaction between Julian day and

intensification: while abundances are similar at the

beginning of June, as the season progresses, the

discrepancy between intensive and extensive landscapes

increases, with fewer Diptera in more intensive land-

scapes at the end of July. Nonetheless, our results

(including those from 2008, which were very different)

indicate that predicting the effects of AI on insect

communities is not a trivial task, both within and among

years.

Temporal patterns in abundance vs. biomass

We observed important variation in Diptera abun-

dance and biomass throughout the sampling period.

This result is not surprising as insect abundances

typically vary in time within and between seasons (e.g.,

Williams 1961, Goulson et al. 2005). Similar to our

results, other studies looking at temporal patterns in

swallow prey availability found large variation during

the season, often reaching a five- to sevenfold difference

between seasonal minima and maxima (Quinney et al.

1986, Hussell and Quinney 1987, Nooker et al. 2005,

Dunn et al. 2011, but see McCarty and Winkler 1999b).

We found a three- to eight-fold difference in Diptera

abundance in 2006 and 2007 between the beginning of

June and the seasonal peak; the extent of this difference

varying among different landscapes. Likewise, biomass

also varied, albeit more linearly during the season for

2006 and 2007, a trend observed in ;50% of site/year

pairs in a previous analysis of Tree Swallow food

abundance at five nesting sites (Dunn et al. 2011).

Considering the large variation in the Diptera

abundance per sample/biomass per sample relationship

(Fig. 2c), it is not surprising that predictive models for

these two response variables differed considerably. But it

leads to an important question, namely: Which variable

best represents the amount of prey available to Tree

Swallows? Total biomass is a parameter associated with

the presence of large-bodied taxa in insect samples.

Dineen et al. (2007) have noted that, in aerial

invertebrate traps, individuals from the suborder Nem-

atocera contributed greatest to abundance, while those

from the suborder Brachycera contributed greatest to

biomass. Diet studies of nestling Tree Swallows suggest

that the majority of prey fed to nestlings are small-

bodied dipterans belonging to the suborder Nematocera,

although they represent less in biomass delivered to

nestlings, and that larger prey (Brachycera) may be

preferred in favorable conditions (Quinney and Ankney

1985, Quinney et al. 1986, McCarty and Winkler 1999a).

Hence, abundance may be a more appropriate indicator

of prey availability in our system, especially at

intensively cultivated sites, but more in-depth investiga-

tion of swallow diet and preference in our study system

may help in pinpointing relevant food availability

parameters in the future.

Seasonal variation in the impact of AI

Several previous studies have emphasized the negative

effects of agricultural intensity on the abundance and

diversity of invertebrate communities (e.g., O’Leske et

al. 1997, Schweiger et al. 2005, Taylor et al. 2006, Evans

et al. 2007, Gruebler et al. 2008). One of the main

mechanisms often cited as responsible for the adverse

effects of intensive agricultural practices is the use of

chemical pesticides and herbicides (Brickle et al. 2000,

Boatman et al. 2004, Taylor et al. 2006), a practice that

is not uniform throughout the bird breeding season. Our

results suggest that the difference in Diptera catches

between farms that are highly agro-intensive and those

that are more extensive has a strong temporal compo-

nent. At day 150, all farms had similar insect abundance

and biomass, but Diptera catches in more intensive

farms tended to be lower than in extensive farms as the

season progressed, in both 2006 and 2007. While

accurate data on timing and quantity of chemical inputs

are difficult to obtain, it is rather safe to assume that

spraying increases as the season progresses and crops

grow.

From an aerial-insectivore perspective, the interaction

between the effect of intensive cultures and time in the

breeding season could lead to an ‘‘ecological trap’’:

During the period when Tree Swallows are selecting

their nest site and laying, they are exposed to similar

conditions (equivalent levels of food availability) re-

gardless of the habitat. However, during fledging and

post-fledging stages, differences in prey abundance

emerge, potentially impacting the success of birds

‘‘trapped’’ in a low-quality habitat. In addition, Porlier

et al. (2009) suggested that early migrants, typically

considered as high-quality individuals, may preferably

January 2013 129DIPTERA ABUNDANCE IN AGROECOSYSTEMS

select intensive habitats due to their proximity to

migration corridors, as most of the intensive agriculture

in our study area is in the westernmost part of the

region, closer to the St. Lawrence River.

Challenge of modeling insect abundance

In comparison with 2006 and 2007, intensive cultures

had a completely opposite effect in 2008, with Diptera

abundance increasing during the season in intensively

cultivated landscapes, while it decreased in extensive

landscapes. Analysis of several weather variables (pre-

cipitation, mean, maximal, and minimal temperatures,

spring snow cover data, degree days accumulation, and

so on) have not provided additional insights as to why

Diptera were more abundant in intensive habitats

during that year (S. Rioux Paquette and M. Belisle,

unpublished data). Detailed taxonomic information,

which is unfortunately unavailable, could be insightful

considering the large variation in dipteran life history

traits, including the number of generations they produce

each summer in given habitats/crops (univoltine vs.

multivoltine), their exposure (e.g., terrestrial vs. aquatic

larvae) and resistance to chemicals (e.g., mutations or

loss of genes involved in pesticide resistance in some

taxa; Huchard et al. 2006), and the occurrence of

swarming behavior, which is common in Diptera

(Downes 1969). Swarms are notoriously difficult to

predict, even in cases where models from weather and

habitat predictors provide otherwise reliable results

(Gruebler et al. 2008). In our case, a single species, the

seedcorn maggot Delia platura (Meigen 1826; family

Anthomyiidae), represented 36% of all brachyceran flies

caught in our system in 2008, and was strongly

associated with maize cultures (L. Laplante, J. Savage,

and M. Belisle, unpublished manuscript). The prolifera-

tion of this saprophagous pest species, known to attack

cereals, is often related to the emergence of other species

that damage roots and facilitate feeding by D. platura

larvae (Beirne 1971, Broatch et al. 2006). D. platura was

shown to be bivoltine in canola crops at similar latitudes

in Alberta, Canada, but variation among years was also

apparent (Broatch et al. 2006), and up to four flights per

year are documented in Ontario (Beirne 1971). Local

abundance of that species can be particularly high at

certain times, even when the population is declining on a

global scale (Higley and Pedigo 1984). Thus, one can

envision the difficulty of modeling such abundance

fluctuations from year to year.

While data from additional years are required before

we can conclude if 2008 results are the exception rather

than a common trend, the majority of empirical studies

around the world support the overall tendency found

here, i.e., the negative effects of AI on insect prey

abundance (e.g., Møller 2001, Ambrosini et al. 2002,

Evans et al. 2007). Interestingly, Tree Swallows in our

system had a much greater overall fledging rate in 2008

than in 2006 and 2007 (Baeta et al. 2012), so higher

dipteran abundance in agro-intensive landscapes that

year could be involved. Reproductive success in Barn

Swallows (Hirundo rustica) is known to be positively

correlated with aerial insect abundance (Møller 2001),

and trophic effects of pesticides have recently been

revealed in House Martins (Delichon urbicum; Poulin et

al. 2011). While our results appear to support the

hypothesis that intensification impacts aerial insecti-

vores through reduced food availability, we wish to

stress that the link between insect abundance and

swallow breeding or post-fledging success was not

investigated here, so caution is warranted regarding

the impacts of AI on bird breeding success through its

influence on prey abundance. However, our results

emphasize that temporal variation should be accounted

for when evaluating landscape effects on arthropod

communities.

Possible implications

Understanding temporal effects is particularly impor-

tant for management strategies because the phenology

of organisms is central in implementing effective

management schemes. For instance, in many grassland

birds nesting at ground level, the most severe impact of

AI is related to the timing of harvest, with earlier cutting

in the breeding season preventing nestlings from

reaching the fledging stage before nests are destroyed

(Green et al. 1997, Perlut et al. 2006). More specifically,

to mitigate these effects on the demography of two

songbird species while maintaining agricultural produc-

tivity, Perlut et al. (2006, 2008) proposed an earlier first

harvest followed by a delayed second harvest in

hayfields, which provides an opportunity for birds to

breed successfully. Similarly, the phenology of pest

insect species is an important aspect of pest management

in crops; the timing of sowing, planting, and harvesting

can be chosen to minimize the vulnerability of crops

during main egg-laying periods by pest insects (Coaker

1987). Our study shows that the effects of intensification

are not uniform during the season, and are potentially

worse in late phases of the swallow breeding season. At

the very end of the breeding season, fledglings may be

particularly vulnerable to these effects, as they become

more adept at finding food and building energy reserves

for migration. This is a significant concern, because food

availability may affect post-fledging survival (as in Lark

Buntings Calamospiza melanocorys; Yackel Adams et al.

2006), which, in turn, is central to songbird population

dynamics (Reid et al. 2011, Streby and Andersen 2011).

However, further aspects need to be investigated before

one could propose tangible strategies to mitigate the

effects of reduced food availability on farmland

insectivore birds. First, direct impacts on swallow

breeding and post-fledging success should be quantified.

Secondly, thorough taxonomic analyses should reveal

the most important species in the swallow diet, and

allow an evaluation of the effects of AI on their life

cycles. For instance, our results show that Diptera

abundance was clearly affected by intensive agriculture

SEBASTIEN RIOUX PAQUETTE ET AL.130 Ecological ApplicationsVol. 23, No. 1

in 2006 and 2007, while biomass did not show the same

pattern, indicating that light-bodied dipterans may

possibly be more vulnerable to the effects of AI. Finally,

obtaining data on the use and accumulation of pesticides

would facilitate the interpretation of results, but this has

been impossible to accomplish so far in our study

system. Ongoing research is conducted with the aim of

dosing pesticides in insects and swallows during the

breeding season.

CONCLUSION

Our analyses defined the seasonal patterns in prey

abundance along an AI gradient and highlighted the

importance of considering the temporal component of

habitat effects on the dynamics of communities.

Although there appears to be a consensus about the

negative effects of AI on insect communities, our results

illustrate that, even with large data sets, predicting the

consequences of agricultural practices is not a trivial

task, both within and among years, and this likely

involves complex ecological interactions. While addi-

tional work is required to assess how seasonal variation

in the effects of AI may affect aerial insectivores,

incorporating the phenology of breeding birds and their

prey into management schemes in agricultural land-

scapes could be an avenue for farmland bird conserva-

tion in the future.

ACKNOWLEDGMENTS

We thank the 40 farm owners who give us access to theirlands. We also thank all students and field assistants who havecontributed, since 2004, in gathering data in our system,collecting insect samples, and sorting and identifying insects.This work was supported by a Fonds Quebecois de laRecherche sur la Nature et les Technologies (FQRNT) newresearcher grant to F. Pelletier and by National Science andEngineering Research Council (NSERC) Discovery grants toM. Belisle, D. Garant, and F. Pelletier, as well as a grant fromthe Canadian Wildlife Service (thanks to J.-P. L. Savard) to M.Belisle.

LITERATURE CITED

Aebischer, N. J. 1991. Twenty years of monitoring invertebratesand weeds in cereal fields in Sussex. Pages 305–331 in L. G.Firbank, N. Carter, J. F. Darbyshire, and G. R. Potts,editors. The ecology of temperate cereal fields. BlackwellScientific Publications, Oxford, UK.

Agriculture and Agri-Food Canada. 1998. Manure, fertilizer,and pesticide management in Canada: Results of the 1995Farm Inputs Management Survey (FIMS). Economic andPolicy Analysis Directorate, Agriculture and Agri-FoodCanada, Ottawa, Canada. http://www3.agr.gc.ca/apps/publicentrale/pub_view-pub_affichage-eng.cfm?publication_id¼1950E&CFID¼1859715&CFTOKEN¼99877024

Ambrosini, R., A. M. Bolzern, L. Canova, S. Arieni, A. P.Møller, and N. Saino. 2002. The distribution and colony sizeof barn swallows in relation to agricultural land use. Journalof Applied Ecology 39:524–534.

Attwood, S. J., M. Maron, A. P. N. House, and C. Zammit.2008. Do arthropod assemblages display globally consistentresponses to intensified agricultural land use and manage-ment? Global Ecology and Biogeography 17:585–599.

Baeta, R., M. Belisle, and D. Garant. 2012. Importance ofbreeding season and maternal investment in studies of sex-

ratio adjustment: a case study using tree swallows. BiologyLetters 8:401–404.

Barton, K. 2011. MuMIn: multi-model inference. R package,version 1.0.0. R Foundation for Statistical Computing,Vienna, Austria. http://CRAN.R-project.org/package¼MuMIn

Batary, P., A. Baldi, D. Kleijn, and T. Tscharntke. 2010.Landscape-moderated biodiversity effects of agri-environ-mental management: a meta-analysis. Proceedings of theRoyal Society Series B 278:1894–1902.

Bates, D., and M. Maechler. 2010. lme4: linear mixed-effectsmodels using S4 classes. R package, version 0.999375-37. RFoundation for Statistical Computing, Vienna, Austria.http://CRAN.R-project.org/package¼lme4

Beirne, B. P. 1971. Pest insects of annual crops in Canada. I.Lepidoptera. II. Diptera. III. Coleoptera. Memoirs of theEntomological Society of Canada 103(Supplement 78):1–124.

Belanger, L., and M. Grenier. 2002. Agriculture intensificationand forest fragmentation in the St. Lawrence valley, Quebec,Canada. Landscape Ecology 17:495–507.

Bengtsson, J., J. Ahnstrom, and A.-C. Weibull. 2005. Theeffects of organic farming on biodiversity and abundance: ameta-analysis. Journal of Applied Ecology 42:261–269.

Benton, T. G., D.M. Bryant, L. Cole, and H. P. Q. Crick. 2002.Linking agricultural practice to insect and bird populations: ahistorical study over three decades. Journal of AppliedEcology 39:673–687.

Benton, T. G., J. A. Vickery, and J. D. Wilson. 2003. Farmlandbiodiversity: is habitat heterogeneity the key? Trends inEcology and Evolution 18:182–188.

Blaxter, K. L., and N. F. Robertson. 1995. From dearth toplenty: the modern revolution in food production. Cam-bridge University Press, Cambridge, UK.

Boatman, N. D., N. W. Brickle, J. D. Hart, T. P. Milsom, A. J.Morris, A. W. A. Murray, K. A. Murray, and P. A.Robertson. 2004. Evidence for the indirect effects ofpesticides on farmland birds. Ibis 146:131–143.

Bohning-Gaese, K., M. L. Taper, and J. H. Brown. 1993. Aredeclines in North American insectivorous songbirds due tocauses on the breeding range? Conservation Biology 7:76–86.

Brickle, N. W., D. G. C. Harper, N. J. Aebischer, and S. H.Cockayne. 2000. Effects of agricultural intensification on thebreeding success of corn buntings Miliaria calandra. Journalof Applied Ecology 37:742–755.

Broatch, J. S., L. M. Dosdall, G. W. Clayton, K. N. Harker,and R.-C. Yang. 2006. Using degree-day and logistic modelsto predict emergence patterns and seasonal flights of thecabbage maggot and the seedcorn maggot (Diptera: Antho-myiidae) in canola. Environmental Entomology 35:1166–1177.

Burnham, K. P., and D. R. Anderson. 2002. Model selectionand multimodel inference: a practical information-theoreticapproach. Second edition. Springer, Berlin, Germany.

Coaker, T. H. 1987. Cultural methods: the crop. Pages 63–69 inA. J. Burn, T. H. Coaker, and C. Jepson, editors. Integratedpest management. Academic Press, London, UK.

Conrad, K. F., M. S. Warren, R. Fox, M. S. Parsons, and I. P.Woiwod. 2006. Rapid declines of common, widespreadBritish moths provide evidence of an insect biodiversitycrisis. Biological Conservation 132:279–291.

Dineen, G., S. S. C. Harrison, and P. S. Giller. 2007. Seasonalanalysis of aquatic and terrestrial invertebrate supply tostreams with grassland and deciduous riparian vegetation.Biology and Environment: Proceedings of the Royal IrishAcademy 107:167–182.

Donald, P. F., R. E. Green, and M. F. Heath. 2001.Agricultural intensification and the collapse of Europe’sfarmland bird populations. Proceedings of the Royal SocietySeries B 268:25–29.

Donald, P. F., F. J. Sanderson, I. J. Burfield, and F. P. J. vanBommel. 2006. Further evidence of continent-wide impacts

January 2013 131DIPTERA ABUNDANCE IN AGROECOSYSTEMS

of agricultural intensification on European farmland birds,1990-2000. Agriculture, Ecosystems and Environment116:189–196.

Douglas, D. J. T., J. A. Vickery, and T. G. Benton. 2010.Variation in arthropod abundance in barley under varyingsowing regimes. Agriculture, Ecosystems and Environment135:127–131.

Downes, J. A. 1969. The swarming and mating flight ofDiptera. Annual Reviews of Entomology 14:271–298.

Duelli, P., M. K. Obrist, and D. R. Schmatz. 1999. Biodiversityevaluation in agricultural landscapes: above-ground insects.Agriculture, Ecosystems and Environment 74:33–64.

Dunn, P. O., D. W. Winkler, L. A. Whittingham, S. J. Hannon,and R. J. Robertson. 2011. A test of the mismatchhypothesis: how is timing of reproduction related to foodabundance in an aerial insectivore? Ecology 92:450–461.

Elliott, N. C., R. W. Kieckhefer, J.-H. Lee, and B. W. French.1998. Influence of within-field and landscape factors on aphidpredator populations in wheat. Landscape Ecology 14:239–252.

ESRI. 2008. ArcGIS version 9.3. Redlands, California, USA.Evans, K. L., J. D. Wilson, and R. B. Bradbury. 2007. Effects

of crop type and aerial invertebrate abundance on foragingBarn Swallows Hirunda rustica. Agriculture, Ecosystems andEnvironment 122:267–273.

Fuentes-Montemayor, E., D. Goulson, and K. J. Park. 2011.Pipistrelle bats and their prey do not benefit from four widelyapplied agri-environment management prescriptions. Biolog-ical Conservation 144:2233–2246.

Gelman, A. 2008. Scaling regression inputs by dividing by twostandard deviations. Statistics in Medicine 27:2865–2873.

Gelman, A., Y.-S. Su, M. Yajima, J. Hill, M. G. Pittau, J.Kerman, and T. Zheng. 2011. arm: data analysis usingregression and multilevel/hierarchical models. R package,version 1.4-07. R Foundation for Statistical Computing,Vienna, Austria. http://CRAN.R-project.org/package¼arm

Ghilain, A., and M. Belisle. 2008. Breeding success of TreeSwallows along a gradient of agricultural intensification.Ecological Applications 18:1140–1154.

Goulson, D., L. C. Derwent, M. E. Hanley, D. W. Dunn, andS. R. Abolins. 2005. Predicting calyptrate fly populationsfrom the weather, and probable consequences of climatechange. Journal of Applied Ecology 42:795–804.

Green, R. E., G. A. Tyler, T. J. Stowe, and A. V. Newton. 1997.A simulation model of the effect of mowing of agriculturalgrassland on the breeding success of the corncrake (Crexcrex). Journal of Zoology 243:81–115.

Gruebler, M. U., M. Morand, and B. Naef-Daenzer. 2008. Apredictive model of the density of airborne insects inagricultural environments. Agriculture, Ecosystems, andEnvironment 123:75–80.

Higley, L. G., and L. P. Pedigo. 1984. Seedcorn maggot(Diptera: Anthomyiidae) population biology and aestivationin central Iowa. Environmental Entomology 13:1436–1442.

Hoye, T. T., and M. C. Forchhammer. 2008. The influence ofweather conditions on the activity of high-arctic arthropodsinferred from long-term observations. BMC Ecology 8:8.

Huchard, E., M. Martinez, H. Alout, E. J. P. Douzery, G.Lutfalla, A. Berthomieu, C. Berticat, M. Raymond, and M.Weill. 2006. Acetylcholinesterase genes within the Diptera:takeover and loss in true flies. Proceedings of the RoyalSociety Series B 273:2595–2604.

Hussell, D. J. T., and T. E. Quinney. 1987. Food abundanceand clutch size of Tree Swallows Tachycineta bicolor. Ibis129:243–258.

Jobin, B., M. Grenier, and P. Laporte. 2005. Using satelliteimagery to assess breeding habitat availability of theendangered loggerhead shrike in Quebec. Biodiversity andConservation 14:81–95.

Krebs, J. R., J. D. Wilson, R. B. Bradbury, and G. M.Siriwardena. 1999. The second silent spring? Nature 400:661–662.

Mazerolle, M. 2011. AICcmodavg: Model selection and multi-model inference based on (Q)AIC(c). R Package, version1.14. R Foundation for Statistical Computing, Vienna,Austria. http://CRAN.R-project.org/package¼AICcmodavg

McCarty, J. P., and D. W. Winkler. 1999a. Foraging ecologyand diet selectivity of Tree Swallows feeding nestlings.Condor 101:246–254.

McCarty, J. P., and D. W. Winkler. 1999b. Relative importanceof environmental variables in determining the growth ofnestling Tree Swallows Tachycineta bicolor. Ibis 141:286–296.

McCracken, J. D. 2008. Are aerial insectivores being ‘buggedout’? BirdWatch Canada 42:4–7.

Mengelkoch, J., G. J. Niemi, and R. R. Regal. 2004. Diet of thenestling Tree Swallow. Condor 106:423–429.

Møller, A. P. 2001. The effect of dairy farming on barn swallowHirundo rustica abundance, distribution and reproduction.Journal of Applied Ecology 38:378–389.

Murphy, M. T. 2003. Avian population trends within theevolving agricultural landscape of eastern and central UnitedStates. Auk 120:20–34.

Nebel, S., and A. Mills, A. J. D. McCracken, and P. D. Taylor.2010. Declines of aerial insectivores in North America followa geographic gradient. Avian Conservation and Ecology 5:1.

Nooker, J. K., P. O. Dunn, and L. A. Whittingham. 2005.Effects of food abundance, weather and female condition onreproduction in tree swallows (Tachycineta bicolor). Auk122:1225–1238.

O’Leske, D. L., R. J. Robel, and K. E. Kemp. 1997. Sweepnet-collected invertebrate biomass from high- and low-inputagricultural fields in Kansas. Wildlife Society Bulletin25:133–138.

O’Rourke, M. E., K. Rienzo-Stack, and A.G. Power. 2011. Amulti-scale, landscape approach to predicting insect popula-tions in agroecosystems. Ecological Applications 21:1782–1791.

Perlut, N. G., A. M. Strong, T. M. Donovan, and N. J.Buckley. 2006. Grassland songbirds in a dynamic manage-ment landscape: behavioral responses and managementstrategies. Ecological Applications 16:2235–2247.

Perlut, N. G., A. M. Strong, T. M. Donovan, and N. J.Buckley. 2008. Regional population viability of grasslandsongbirds: effects of agricultural management. BiologicalConservation 141:3139–3151.

Porlier, M., M. Belisle, and D. Garant. 2009. Non-randomdistribution of individual genetic diversity along an environ-mental gradient. Philosophical Transactions of the RoyalSociety Series B 364:1543–1554.

Poulin, B., G. Lefebvre, and L. Paz. 2011. Red flag for greenspraying: adverse trophic effects of Bti on breeding birds.Journal of Applied Ecology 47:884–889.

Quinney, T. E., and C. D. Ankney. 1985. Prey size selection byTree Swallows. Auk 102:245–250.

Quinney, T. E., D. J. T. Hussell, and C. D. Ankney. 1986.Sources of variation in growth of Tree Swallows. Auk103:389–400.

R Development Core Team. 2011. R: A language andenvironment for statistical computing. R Foundation forStatistical Computing, Vienna, Austria.

Reid, J. M., E. M. Bignal, S. Bignal, M. I. Bogdanova, P.Monaghan, and D. I. McCracken. 2011. Diagnosing thetiming of demographic bottlenecks: sub-adult survival in red-billed choughs. Journal of Applied Ecology 48:797–805.

Robinson, R. A., and W. J. Sutherland. 2002. Post-war changesin arable farming and biodiversity in Great Britain. Journalof Applied Ecology 39:157–176.

Robinson, S. K., F. R. Thompson III, T. M. Donovan, D. R.Whitehead, and J. Faaborg. 1995. Regional forest fragmen-

SEBASTIEN RIOUX PAQUETTE ET AL.132 Ecological ApplicationsVol. 23, No. 1

tation and the nesting success of migratory birds. Science267:1987–1990.

Sauer, J. R., G. W. Pendleton, and B. G. Peterjohn. 1996.Evaluating causes of population change in North Americaninsectivorous songbirds. Conservation Biology 10:465–478.

Schielzeth, H. 2010. Simple means to improve the interpret-ability of regression coefficients. Methods in Ecology andEvolution 1:103–113.

Schweiger, O., et al. 2005. Quantifying the impact ofenvironmental factors on arthropod communities in agricul-tural landscapes across organizational levels and spatialscales. Journal of Applied Ecology 42:1129–1139.

Shortall, C. R., A. Moore, E. Smith, M. J. Hall, I. P. Woiwod,and R. Harrington. 2009. Long-term changes in theabundance of flying insects. Insect Conservation andDiversity 2:251–260.

Sileshi, G., G. Hailu, and G. I. Nyadzi. 2009. Traditionaloccupancy-abundance models are inadequate for zero-inflated ecological count data. Ecological Modelling220:1764–1775.

Statistics Canada. 2006. Census of agriculture. StatisticsCanada, Ottawa, Ontario, Canada. www.statcan.gc.ca/ca-ra2006/index-eng.htm

Streby, H. M., and D. E. Andersen. 2011. Seasonal productivityin a population of migratory songbirds: why nest data are notenough. Ecosphere 2:78.

Taylor, R. L., B. D. Maxwell, and R. J. Boik. 2006. Indirecteffects of herbicide on bird food resources and beneficial

arthropods. Agriculture, Ecosystems and Environment116:157–164.

Tscharntke, T., A. M. Klein, A. Kruess, I. Steffan-Dewenter,and C. Thies. 2005. Landscape perspectives on agricultural

intensification and biodiversity: ecosystem service manage-ment. Ecology Letters 8:857–874.

Vickery, J. A., J. R. Tallowin, R. E. Feber, E. J. Asteraki, P. W.Atkinson, R. J. Fuller, and V. K. Brown. 2001. The

management of lowland neutral grasslands in Britain: effectsof agricultural practices on birds and their food resources.Journal of Applied Ecology 38:647–664.

Wickramasinghe, L. P., S. Harris, G. Jones, and N. V. Jennings.

2004. Abundance and species richness of nocturnal insects onorganic and conventional farms: effects of agriculturalintensification on bat foraging. Conservation Biology

18:1283–1292.

Williams, C. B. 1961. Studies in the effect of weather conditionson the activity and abundance of insect populations.Philosophical Transactions of the Royal Society of London

B 244:331–378.

Yackel Adams, A. A., S. K. Skagen, and J. A. Savidge. 2006.Modeling post-fledging survival of lark buntings in response

to ecological and biological factors. Ecology 87:178–188.

Zuur, A. F., E. N. Ieno, N. J. Walker, A. A. Savaliev, andG. M. Smith. 2009. Mixed effects models and extensions inecology with R. Springer, New York, New York, USA.

SUPPLEMENTAL MATERIAL

Appendix A

Distribution of the dates of laying, hatching, and fledging of the first egg/nestling in 929 Tree Swallow clutches among 400 nestboxes in the study area for three years (2006–2008) (Ecological Archives A023-010-A1).

Appendix B

Preliminary analysis of the diet of Tree Swallow nestlings in a system of nest boxes located in southeastern Quebec, Canada,from collected boluses (Ecological Archives A023-010-A2).

Appendix C

Model selection summary regarding Diptera abundance and biomass measured on 40 farms located in southeastern Quebec,Canada, 2006–2008 (Ecological Archives A023-010-A3).

Appendix D

Model-averaged, standardized parameter values regarding Diptera abundance and biomass measured on 40 farms located insoutheastern Quebec, Canada, 2006–2008 (Ecological Archives A023-010-A4).

Appendix E

Model selection summary regarding Diptera abundance and biomass measured on 40 farms located in southeastern Quebec,Canada, for three years separately (2006, 2007, and 2008) (Ecological Archives A023-010-A5).

Appendix F

Goodness-of-fit of linear mixed models computed to describe Diptera abundance and biomass patterns on 40 farms located insoutheastern Quebec, Canada, in 2006, 2007, and 2008 (Ecological Archives A023-010-A6).

Appendix G

Model-averaged, standardized parameter values regarding Diptera abundance measured on 40 farms located in southeasternQuebec, Canada, for three years separately (2006, 2007, and 2008) (Ecological Archives A023-010-A7).

Appendix H

Model-averaged, standardized parameter values regarding Diptera biomass measured on 40 farms located in southeasternQuebec, Canada, for three years separately (2006, 2007, and 2008) (Ecological Archives A023-010-A8).

January 2013 133DIPTERA ABUNDANCE IN AGROECOSYSTEMS