1793
population.
1794 1795 Résumé 1796De nombreuses études ont démontrées l’effet positif de la diversification végétale à l’échelle
1797
d’une parcelle. A l’échelle du paysage, la proportion d’habitats semi-naturels a également été
1798
prouvé très positif pour la présence des ennemis naturels au champ et leur biocontrôle.
1799
Cependant, peu d’études ont évalué l’effet de la diversification végétale cultivée à l’échelle du
1800
paysage sur les ennemis naturels et leur biocontrôle. Afin de tester l’hypothèse qu’une
1801
succession culturale diversifiée pourrait augmenter l’abondance des ennemis naturels, leur
1802
biocontrôle et leurs mouvements (spillover), nous avons testé en conditions contrôlées sous
1803
serre l’effet de deux types de successions culturale sur l’ennemi naturel Nesidiocoris tenuis
1804
Reuter (Hémiptère : Miridae). Les successions culturales testées étaient les suivantes :
1805
succession différenciée dans le temps (45 jours) et l’espace (a) d’un type de culture (soit tomate,
1806
soit courge et soit soja), ou (b) des trois cultures assemblées, avec la tomate en première culture,
1807
courge en seconde culture et soja en troisième culture. Nos résultats ont montré que (i) la
1808
croissance de population du prédateur ainsi que la prédation d’œufs d’Ephestia kuehniella
1809
Zeller (Lépidoptère : Pyrale) étaient moins importants en polyculture qu’en monocultures de
1810
tomate et courge, mais (ii) plus importants que sur monoculture de soja. Ces résultats montrent
1811
que la nature des cultures plus que leur diversité est un point crucial dans la préservation des
1812
prédateurs en champ, surtout lorsqu’il s’agit d’un prédateur hautement dépendant des plantes,
1813
car se nourrissant de leur sève comme celui que nous avons étudié ici. La présence de soja, sur
1814
lequel le prédateur ne se développait pas correctement, en fin de succession en polyculture a en
1815
effet réduit l’efficacité global du système polycultural. De manière intéressante, à la fin de la
1816
période d’expérimentation, lorsque la population de prédateur s’est transférée sur la culture
1817
soja, le nombre d’œufs prédaté par individu et par jour était plus important qu’en monocultures
1818
combinées. Ce dernier résultat suggère que des plantes non adaptées au développement du
1819
prédateur mais adjacente aux cultures qui le sont pourraient bénéficier du spillover des ennemis
1820
naturels et augmenter ainsi le potentiel de biocontrôle dans ces cultures.
1821
Mots-clés : polyculture, monoculture, succession culturale, prédation, spillover
1822 1823
150 1824
1825 1826
Effect of crop diversification on predation efficiency and population dynamics of the
1827
mirid bug Nesidiocoris tenuis.
1828
Eva Thominea, Emma Jeavonsb, Adrien Ruschc, Philippe Bearez a, Nicolas Desneuxa* 1829
a: INRA (French National Institute for Agricultural Research), Université Côte D’Azur, CNRS,
1830
UMR 1355-7254, Institut Sophia Agrobiotech, 400 Route des Chappes, 06903 Sophia Antipolis,
1831
France
1832
b: CNRS 6553 Ecosystems, Biodiversity and Evolution (ECOBIO), University Rennes 1, 35042
1833
Rennes, France
1834
c: INRA, ISVV, Univ. Bordeaux, Bordeaux Sciences Agro, UMR SAVE, F-33883 Villenave
1835
d'Ornon, France
1836
*corresponding author: Nicolas Desneux: nicolas.desneux@inra.fr, +33492386427
1837 1838
151
Key messages
1839 1840
x Underlying mechanisms of landscape crop diversity effect on natural enemies has
1841
been scarcely described.
1842
x We hypothesized that spatio-temporal succession of diversified crops enhances
1843
biocontrol service by supporting the spillover of predators among crops.
1844
x Single crop effects on the omnivorous predator were cumulated in the polycultural
1845
system.
1846
x Our results suggest that non-host adjacent crops, placed in a crop succession, can
1847
benefit from the presence of host plants where natural enemy populations have been
1848
increased.
1849 1850
152
A
Abstract
1851
A large body of evidence has shown the positive impact of plant species richness on the field
1852
scale. The proportion of semi-natural habitats in the landscape has also impacted natural enemy
1853
communities and biological control services. However, very few studies have assessed the
1854
effect of crop diversity in the landscape on natural enemy performances and pest control. In
1855
order to test the hypothesis that crop diversity could increase natural enemy development and
1856
performance, we examined the underlying mechanisms modulating the effect of two types of
1857
crop successions, i.e. multiple-crop succession (tomato, squash and soybean) and mono-crop
1858
succession (each crops alone), on population dynamics, predation capacity and spillover of
1859
Nesidiocoris tenuis Reuter (Hemiptera: Miridae), the mirid bug, in a greenhouse experiment.
1860
We found that (i) polyculture supported lower population growth of N. tenuis and biological
1861
control of Ephestia kuehniella Zeller (Lepidoptera: Pyralidae) eggs compared to tomato and
1862
squash monocultures, but that on the other hand (ii) the predator performed better on
1863
polyculture than on soybean monoculture. These results revealed that crop identity within the
1864
succession was a major factor in clarifying population dynamics and biological control. We
1865
found that the presence of soybean crop Glycine max L. (Fabales: Fabaceae) in the polyculture
1866
treatment reduced the population dynamics of the mirid bug but increased biocontrol. This
1867
result suggests that non-host adjacent crops in a cultural succession can benefit from the
1868
presence of host plants where the natural enemy population is increased.
1869
Key words: habitat management, crop succession, conservation biological control, predation,
1870
spillover
1871
153
IIntroduction
1873
Modern industrial agriculture has resulted in considerable negative environmental impacts,
1874
such as side effects of pesticides on beneficial arthropods, habitat fragmentation, diversity loss
1875
etc., which threaten sustainability of most food production systems (Kareiva 1987; Desneux et
1876
al. 2007; Lu et al. 2012; Jonsson et al. 2014). Ecological intensification offers a way of reducing
1877
these environmental impacts by improving ecological processes in agroecosystems and limiting
1878
agrochemical dependency (Cardinale et al. 2006; Letourneau et al. 2015). Among the
1879
approaches used in ecological intensification, conservation biological control, based on the use
1880
of locally present natural enemies to reduce pest populations, is an important ecosystem service
1881
that can help to reduce pesticide use and crop damage (Parolin et al. 2012; Zhao et al. 2017;
1882
Karp et al. 2018; Perovic et al. 2018).
1883
Key resources such as pollen, nectar, nesting sites and alternative hosts and preys within
1884
agricultural landscapes are known to be major drivers of natural enemy population dynamics
1885
(Gurr et al., 2017; Perovic et al. 2018). Enhancing plant diversity to maintain multiple
1886
resources, either on the field or on the landscape level, can possibly boost natural enemy
1887
populations and reduce pest infestation levels (Letourneau et al. 2011; Bianchi et al. 2006;
1888
Rusch et al., 2016; Karp et al., 2018; Gurr et al 2017). A diversity of resources to target all
1889
natural enemy life stages and the spatiotemporal continuity of those resources in agricultural
1890
landscapes are crucial elements to consider to maintain natural enemy populations throughout
1891
the year (Schellhorn et al., 2015). At the landscape level, enhancing plant diversity often goes
1892
through increasing the proportion of semi-natural habitats (i.e. habitat complexity) (Rusch et
1893
al., 2016; Karp et al., 2018; Gurr et al 2017)
1894
However, increasing plant diversity through semi-natural habitats is costly in terms of space
1895
and management for farmers. In a context of increasing food demand, extensive growth of urban
1896
surfaces and reduced agricultural intensification, increasing semi-natural areas in agricultural
154
landscape seems difficult to implement (Burton et al. 2008; Brewer and Goodell 2012; Bianchi
1898
et al. 2006). Playing on crop diversity as a way to provide key resources to natural enemies is
1899
a complementary approach that could be easier to implement. Indeed, this approach would
1900
provide direct value to farmers, and thus may incite them to adopt such systems (Vasseur et al.,
1901
2013). Crops can provide high amounts of a specific resource (nectar, pollen or alternative
1902
hosts/preys), but only during a brief period in conventional agricultural systems (Rand et al.,
1903
2006; Tscharntke et al., 2005). To maintain natural enemy populations with cropping
1904
diversification systems, crops have to be assembled in order to provide complementary
1905
resources in a continuous way, thus avoiding food chain interruptions (Vasseur et al. 2012;
1906
Schellhorn et al. 2015). A thorough understanding of the quality and temporality of the
1907
resources provided by each crop is crucial to be able to associate crops in a polycultural system
1908
targeting natural enemies. Of course, not all natural enemies have the same requirements,
1909
therefore some crops may be less suitable than others for different natural enemies. Associating
1910
diverse crops in a continuous polyculture could allow that, for a given natural enemy, a less
1911
suitable crop is temporally and spatially surrounded by suitable crops, making the system more
1912
resilient than a monoculture of the less suitable crop for the natural enemy. Therefore, we
1913
expect that crop diversification in a continuous system could support natural enemy populations
1914
more than simplified cropping systems, and that this population enhancement could lead to
1915
higher biological control. In such systems, natural enemies should be able to spill-over (i.e.
1916
transfer) from one crop to another. Nevertheless, many mechanisms underlying the effect of 1917
crop diversification on the biological control of pests are lacking.
1918
Predatory bugs are natural enemies often used in biological control programs, and is the main
1919
pest control strategy currently used in southeast Spain on tomato and sweet pepper crops (Pérez-
1920
Hedo and Urbaneja, 2015a). Predatory bugs are known to be omnivorous and generalist
1921
predators feeding on different types of preys and on plant materials. Many studies have shown
155
that mirid bugs can be found on many different types of host plants, mainly Solanaceae,
1923
Asteraceae, Cucurbitaceae, Fabaceae and Pedaliaceae (Naselli et al., 2017; Biondi et al., 2016;
1924
Sanchez et al., 2003). Nesidiocoris tenuis (Reuter) (Hemiptera : Miridae) is a mirid bug largely
1925
used in greenhouses in South of Spain as it has the capacity to feed on many different types of
1926
key pests. It feeds on both plant sap, pollen, nectar and soft bodied insects or eggs (Molla et al.,
1927
2014; Arno et al. 2010; Urbaneja et al., 2009; Calvo et al., 2009), thus making it an ideal
1928
candidate to investigate the effect of plant diversification on biological control. It has
1929
particularly been studied for its capacity to control the tomato leaf miner Tuta absoluta
1930
(Meyrick) (Lepidoptera : Gelechiidae) (Urbaneja et al., 2009) and the whitefly Bemisia tabaci
1931
(Gennadius) (Hemiptera : Aleyrodidae) (Calvo and Urbaneja, 2004), but also aphids on pepper
1932
plants (Perez-Hedo and Urbaneja, 2015). Additionally, N. tenuis was of particular interest for
1933
our study as its benefits for crops were proved to be also indirect through volatiles elicitation
1934
and consequent parasitoid mobilization and pests disruption (Pérez-Hedo et al., 2015b). Finally,
1935
despite the propensity of the mirid bug to cause necrotic rings, new studies have found that by
1936
manipulating tomato varieties or temperatures, the damages caused by the predator could be
1937
reduced and therefore making it a good candidate for biocontrol in greenhouses (Siscaro et al.,
1938
2019). N. tenuis omnivorous trait makes the insect dependent on the plant species and therefore,
1939
makes it suitable to examine the question of plant diversity and succession impact on natural
1940
enemies.
1941
This study aims at analyzing the mechanisms linked to the effect of diversified crop richness
1942
and continuous crop succession on the population growth, the predation capacity and the
1943
spillover of the mirid bug Nesidiocoris tenuis Reuter (Hemiptera: Miridae). We expected, in
1944
particular, that increasing crop species richness over space and time would lead to: (i) higher
1945
population growth of the natural enemy, (ii) enhance predation efficiency of N. tenuis, (iii)
1946
show a higher spillover of predators compared to the monoculture system. These effects would
156
be linked to higher availability and diversity of food sources in the polycultures compared to
1948
the monocultures.
1949
M
Material and methods
1950
Studied organisms
1951
Predatory mirid bug
1952
The predator studied in this experiment is the mirid bug N. tenuis Reuter (Hemiptera: Miridae).
1953
The mirid bug colony was reared on tomato plants under laboratory conditions (25 ± 3°C; LD
1954
15:9; 50 ± 10% RH). The colony was regularly fed with sterilized E. kuehniella Zeller eggs
1955
(Biotop ®, France), honey diluted in water (1/4) and pollen grains from diverse type of flowers
1956
obtained with usual beekeeping extraction.
1957
Pest
1958
The pest selected in this experiment was a substitute prey: sterilized E. kuehniella Zeller
1959
(Lepidoptera: Miridae) eggs. This substitute pest is well accepted in literature and is easy to
1960
obtain as a food source for many natural enemies. Additionally, other Pyralidae are important
1961
pest species in many agricultural systems and therefore E. kuehniella is a good representative
1962
of this category of pests. Finally, as only sterilized eggs were taken for the experiment, there
1963
was no pest population dynamics in order to simplify the trophic system and focus on the direct
1964
effects of plant diversity and succession on the natural enemy.
1965
Plants
1966
Three different plants were chosen for the design: tomato (Solanum lycopersicum L.
1967
(Solanacea) - Nano), squash (Cucurbita moschata D. (Cucurbitacea) - Butternut) and soybean
1968
(Glycine max L. (Fabaceae) - Merrill). These plants were chosen because they are
1969
complementary in terms of: (i) family and (ii) provision of nectar and pollen (see Tab. 1). Pre-
1970
tests were done in order to confirm that the mirid bug is able to grow on the different chosen
1971
plants. During the experiment, all plants reached the flowering stage, allowing the predatory
1972
bug to feed on pollen and nectar sources when available and if necessary, in addition to sap
157
collected mainly on the apical vegetative sprouts (Siscaro et al., 2019) and to Ephestia’s eggs
1974
available on predation cards.
1975
Experimental design
1976
The experiment was carried out under greenhouse conditions from 17th May to 27st July 2017 1977
at the INRA site in Sophia Antipolis. The temperature, as well as the relative humidity, were
1978
stabilized using cooling and fogging systems (25 ± 10°C; LD 15:9 ± 1h; 70 +20-50% RH). The
1979
experimental greenhouse was composed of four hermetically separated compartments, each
1980
measuring 6*6 m. In each compartment, four double rows of metal poles measuring each 5m
1981
length were placed as a support for plants. A nutritive solution composed of NPK (165:117:225)
1982
and other trace elements was provided four times a day to all the plants in the experiment.
1983
Treatments
1984
To look into the effects of crop diversity on N. tenuis, the succession of three crops, hereafter
1985
named “polyculture”, was compared to the successions of the same crop for each individual
1986
crop separately, hereafter named “monocultures” (Table 1 and Fig. 1). Consequently, there were
1987
four treatments: the tomato monoculture, the squash monoculture, the soybean monoculture
1988
and the polyculture. Each treatment was replicated four times, once in each compartment. A
1989
replicate consisted in a double row of plants which was isolated from the others by means of a
1990
net tunnel. In order to assess the question of plant succession and natural enemy spillover, each
1991
double row of plants was divided into three equal parts representing a unit of space and time
1992
called “patch type” in which 8 plants were placed: A, B and C. In the polyculture treatment, the
1993
plants in patch type A were tomato plants, squash plants were in patch type B and soybean
1994
plants in patch type C. The order of plant succession was chosen regarding usual plant
1995
succession in continental plant cropping systems. In the monoculture treatments, all patch types
1996
contained the same crop. The patch types of plants were added gradually, every 15 days.
1997
Nevertheless, in order to assess the effect of habitat disturbance in the system, each plant patch
1998
type was cut 25 days after its implementation, and the crop residues were left to allow potential
158
offspring to develop. To sum up, after implementing plants in patch type A, on day 15 patch
2000
type B was added; then on day 25, patch type A was cut, on day 30 patch type C was added and
2001
on day 40 patch type B was cut. Thanks to this technique, the insects always had 10 days to
2002
switch from one patch type to another.
2003
Predator population dynamics and spillover between the patch types
2004
On day 1, 10 young N. tenuis females and 10 young N. tenuis males were placed in each
2005
treatment. The N. tenuis populations were sampled in each tunnel every five days. Insects were
2006
counted on the living plants, on plant residues as well as in the environment nearby. In order to
2007
avoid any overlapping during the insect sampling, each individual was lifted out carefully using
2008
a mouth aspirator for counting in one patch and then put back in the middle of the same patch
2009
at the end of the counting. All instars were noted, i.e. nymphal instars 1, 2, 3, 4 and 5 and male
2010
and female adults. To assess the spill-over of the predators, evaluated as the movement of the
2011
predatory bugs from one patch type to another, during the sampling the patch types were
2012
physically separated by a net in order to avoid any movement of the predators from one patch
2013
to another induced by the disturbance of the sampling. The position of the populations
2014
individuals were then recorded in each patch type. Population spill-over was assessed by
2015
calculating the percentage of the population present in a patch type, compared to the previous
2016
patch type, based on the formula below:
2017
ܰݑܾ݉݁ݎ݂݅݊ݏ݁ܿݐݏ݅݊ܲܽݐ݄ܿݐݕ݁ܺܽݐݐ݅݉݁ݐ כ ͳͲͲ ܰݑܾ݉݁ݎ݂݅݊ݏ݁ܿݐݏ݅݊ܲܽݐ݄ܿݐݕ݁ሺܺ െ ͳሻܽݐݐ݅݉݁ݐ
2018
The spill-over was also assessed through the comparison of the population of N. tenuis between
2019
the patch types within a treatment. Indeed, patch types represent a unit of space, and as such
2020
provide a good basis to measure the movement intensity of the population. We assumed that
2021
population abundance may vary between the patches type depending on the plants available in
2022
the system.
159
Predation performance
2024
Predation cards made with E. kuehniella sterilized eggs were used to estimate N. tenuis
2025
predation performances (Winqvist et al., 2011). The amount of eggs exposed to mirid bugs,
2026
around 600 per predation card, was based on the results from preliminary experiments. In the
2027
greenhouse, predation cards were disposed homogenously on the living plants, and were
2028
replaced every two days to avoid dehydration or total predation. Eight cards per patch type were
2029
installed, firstly every two days and when the second generation of N. tenuis appeared, i.e. from
2030
day 12 to the end, 16 cards were installed. As predated eggs and dehydrated eggs were difficult
2031
to distinguish, control cards with eggs inaccessible to the predators, protected with a small bag
2032
composed by nylon mesh, were also placed in the tunnels in order to measure a mean
2033
dehydration rate per compartment on each date. The number of predated eggs per patch type
2034
was estimated every two days by counting damaged eggs using a binocular and the mean
2035
dehydration rate was deducted. The mean number of predated eggs per insect per day was also
2036
recorded.
2037
Statistical analysis
2038
The effect of the different treatments on N. tenuis population and predation efficacy, which
2039
followed a Poisson error distribution, was analyzed using Generalized Estimating Equations
2040
(GEE) with the patch type, the date, separately and in two to three sided interactions as
2041
explanatory variables. We specified a first-order auto-regressive correlation structure “ar1”,
2042
which is based on the assumption that observations close in time are much more correlated than
2043
observations further apart. The GEE model was simplified using a backward stepwise method.
2044
The post-hoc test “lsmeans” (Length, 2016) was used to compare every treatment to each other.
2045
The spillover of the predator population, which followed a Poisson error distribution, was
2046
analyzed with a Generalized Linear Model (GLM) with patches considered as a unit of space
2047
and time. The variable used to explain was the total number of predators and the explanatory
2048
variable was the treatment and the patch type in interaction.
160
All statistical analyses were carried out with the R software v3.5.1 using the geepack package
2050
(Hojsgaard et al., 2006), the lsmeans package (Lenth, 2016), and the multcomp package
2051 (Horthorn et al., 2008). 2052
R
Results
2053 Population dynamics 2054The number of N. tenuis was significantly lower in the soybean monoculture in every patch
2055
type (Fig. 2, Table S1; soybean – squash: z.ratio: -15.1, P < 0.001; soybean – tomato: z.ratio: -
2056
13.5, P < 0.001; soybean – polyculture: z.ratio: -11.6, P < 0.001) with a population of initial
2057
adults dropping prematurely in soybean. The number of N. tenuis was always significantly
2058
lower in the soybean monoculture in each patch type (Table S2; all P < 0.001). In tomato and
2059
squash monoculture, abundance of N. tenuis was equal to the polyculture in patch type A and
2060
B (Table S3; all P > 0.05) whereas in patch type C numbers were significantly lower in the
2061
polyculture (Fig. 2; squash – polyculture in patch type C: z.ratio: 5.2, P < 0.001; tomato –
2062
polyculture in patch type C: z.ratio: 4.8, P < 0.001).
2063
Population spillover
2064
The population spill-over was significant for each treatment as patch type had a significant
2065
impact on the number of N. tenuis (Table S3; treatment: χ2
1: 854.1, P < 0.001; patch type: χ22: 2066
2024.95, P < 0.001). The spillover, measured as the percentage of population transfer from on
2067
patch type to another, could not be calculated on the soybean monoculture because there were
2068
too few insects. In the tomato and the squash monocultures, the amount of population transfer