Natural and released inoculum levels of entomopathogenic fungal biocontrol agents in soil in relation to risk assessment and in accordance with EU regulations

Abstract

Entomopathogenic fungi being developed as biological control agents (BCAs) may have the potential to spread and become established in the environment. For registration purposes, the risks concerning their persistence have to be evaluated according to EU legislation which requires the decline of BCAs to acceptable background levels unless related risks are acceptable. In order to deal with this requirement, applicants of a BCA need to give information on its persistence and natural background levels. For risk assessors and registration authorities guidance on how to evaluate data on natural background levels of indigenously occurring species (=same species as the introduced BCA) is needed. For this purpose, an overview is presented on background levels of some indigenous fungi as well as persistence data of some applied fungal BCAs. Data were restricted to commerical species. It was found that for the species Metarhizium anisopliae, Beauveria bassiana and Beauveria brongniartii natural densities were relatively low and introduced strains of these fungi decreased gradually in time. Many factors were found in the literature, such as intrinsic, edaphic, biotic, climatic, and cultural factors, that could explain this decline.

Keywords:

• data requirements
• registration
• Council Directive 91/414/EEC
• Council Directive 2001/36/EC
• Council Directive 2005/25/EC
• biological control agents
• entomopathogenic fungi
• occurrence and persistence in soil
• natural background concentrations

Introduction

Entomopathogenic fungi (EPF) contribute to the natural regulation of insect, tick and mite populations (Butt, Jackson, and Magan 2001; Kleespies, Huger, and Zimmermann 2008). Over 750 different species have been identified to date. These are cosmopolitan organisms which have been isolated from soils and infected insects from around the world as evidenced by notable culture collections such as the USDA-ARS culture collection of insect-pathogenic fungi (http://arsef.fpsnl.cornell.edu/mycology/ARSEF_Culture_Collection.html). The natural occurrence of these fungi is also documented by numerous researchers from around the world as summarised in the excellent reviews by Zimmermann (2007a, 2007b, 2008). Several species of EPF have been developed as benign alternatives to synthetic chemical insecticides that have been withdrawn because of the risks they pose to humans and the environment and to which pests have developed resistance (Butt et al. 2001). Commercial or commercially viable species of EPF include species like Metarhizium anisopliae, Beauveria bassiana, Beauveria brongniartii, Lecanicillium spp. (formerly Verticillium spp.) and Paecilomyces (now in genus Isaria) fumosoroseus. Examples of registered EPF-based products are provided by Copping (2004) and De Faria and Wraight (2007).

Risk assessment poses one of the major hurdles in the commercialisation of EPF. The introduction of plant protection products of chemical and biological origin onto the European market is governed by Council Directive 91/414/EEC1 (European Commission 1991) of the European Commission. Each product needs to be evaluated according to the Uniform Principles of Annex VI (European Commission 2005). Risk assessment of microbial biological control agents (BCAs) is a relatively recent process compared with the risk assessment of chemical pesticides. Unfortunately, some tools for risk assessment of chemical pesticides are not appropriate for fungal BCAs and can pose problems in the registration of these organisms as plant protection products. Attempts have been made to deal with this issue. For example, the REBECA (http://www.rebeca-net.de) consortium (REBECA 2006–2007) investigated ways to simplify the registration procedure for BCAs as a whole. The RAFBCA (http://www.rafbca.com) consortium (RAFBCA 2001–2004) investigated the risks posed by metabolites of fungal BCAs to humans and the environment. They developed methods and tools to determine the risks posed by these metabolites and showed that the metabolites of the few fungi investigated did not pose a risk. Mensink and Scheepmaker (2007) developed a decision tree for the risk assessment of microbial BCAs and concluded that risk assessors needed more tools to accelerate the risk assessment process. In contrast with chemical pesticides, guidance documents are not available. This review is intended to help develop such a tool.

Improved communication, coordination and cooperation between the key stakeholders will accelerate the process in determining which methods and tools are essential for risk assessment of microbial BCAs. Furthermore, in depth analysis of existing data is required not only in helping develop the tools for risk assessment but also modification of data requirements especially where it is shown that risks are acceptable. The latter could help avoid industry conducting certain studies (i.e., prepare waivers or statements instead) and accelerate the evaluation process.

In this review we focus on EC Directives 91/414 (European Commission 1991), 2001/36/EC requirements (European Commission 2001) and the Uniform principles of micro-organisms Directive 2005/25/EC (European Commission 2005) on the fate and behaviour of EPF as plant protection products in soil. Regulators request these data because BCAs are considered to have the potential to spread and become established in the environment, and to increase. The authorities have formulated data requirements to investigate any long-term non-target effects of these agents. Possible effects include (1) competitive displacement of non-target micro-organisms in the resident soil community, (2) toxicity of biologically active metabolites, (3) pathogenicity to non-target organisms, and (4) allergenicity to humans or other animals (Cook et al. 1996; Goettel, Hajek, Siegel, and Evans 2001).

Of particular concern in the Uniform Principles for micro-organisms is point 2.7.7 on persistence of the inoculum in the soil. This point states that: ‘No authorisation shall be granted if it can be expected that the micro-organism and/or its possible relevant metabolites/toxins will persist in the environment in concentrations considerably higher than the natural background levels, taking into account repeated applications over the years, unless a robust risk assessment indicates that the risks from accumulated plateau concentrations are acceptable’. The Uniform Principles do not define how to establish the background concentration of a particular species. Clearly, the time span of persistence above background levels is prone to a wide array of interpretations. Also, it is not clear how to interpret the phrase ‘considerably higher than the natural background levels’. Can we allow concentrations to be, for instance, a factor 5, 10 or 100 higher than the background concentration? The ultimate decision has to be made by risk managers, but first the relation between the concentration levels and possible non-target effects need to be clarified and evaluated by risk assessors.

This review provides the first detailed analysis of natural background concentrations of three species of EPF and the persistence of introduced inocula. Reference is made to other EPF species (Lecanicillium and Paecilomyces) where it is felt it may help reinforce or clarify a specific point. Factors are described which could explain the decline in inoculum levels. We also discuss the significance of our findings for possible rephrasing of data requirements.

Background levels of native EPF

In order to be able to assess whether a BCA is persistent, risk assessors need to have information about natural background levels of the indigenous species (preferable the same species and the same strain). Background levels of indigenous BCAs may follow the population dynamics of the insect host population as illustrated by Rath, Worledge, Koen, and Rowe (1995b) where an increase of the host Adoryphorus couloni was closely followed by an increase in levels of the pathogen, M. anisopliae. Similarly, Strasser (1999) observed that in pastures, cockchafer (Melolontha melolontha) numbers increased between May and September followed by an increase in the pathogen B. brongniartii. Fluctuations in the background levels also depend on a range of other factors including the climate, season, soil type, biotope (e.g., forest or arable field), crop type, and agronomic practices (see section ‘factors explaining the decline in inoculum levels’).

Methods to quantify EPF in the field

Before reviewing natural levels of inocula, it is important to examine the methods used to determine the levels of fungal BCAs, whether natural or introduced.

Selective media

The most common methods of determining levels of inoculum is based on isolation using selective media (Veen and Ferron 1966; Beilharz, Parberry, and Swart 1982; Baath 1991) and insect baiting methods (Zimmermann 1986). Selective media can be viewed as a semi-quantitative method. Most selective media contain either a fungicide and/or antibiotics which encourage growth of entomogenous fungi and discourage growth of saprophytic fungi and bacteria. Fastidious strains of EPF may fail to establish on the selective media thus giving a false picture of the species diversity and density. These methods are reviewed in more detail by Butt and Goettel (2000) and Lacey (1997).

Insect baiting

For insect baiting, larvae of Galleria mellonella are preferentially used because they are highly susceptible to fungal infection but other insect species have also been used such as Tenebrio molitor (Vänninen, Tyni-Juslin, and Hokkanen 2000; Pilz, Wegensteiner, and Keller 2008) and Delia floralis (Klingen, Eilenberg, and Meadow 2002). The insect bait method has several drawbacks. Firstly, it is a qualitative method which determines whether a fungus is present or not. It will only detect strains that are compatible with the target insect thus under-estimating the range of species/strains in the soil. For example, D. floralis was more effective in isolating Tolypocladium cylindrosporum than G. mellonella while the latter was better in isolating M. anisopliae and B. bassiana (Klingen et al. 2002). In the case of co-infections, usually one fungal species dominates (Wang, Li, and Butt 2002), thereby underestimating the occurrence of the subdominant species. The temperature at which insect baiting is performed may influence which species are isolated. For example, Sookar, Bhagwant, and Awuor Ouna (2008) found that B. bassiana was isolated more frequently at the lowest incubation temperature (15°C) while M. anisopliae isolates were recovered more frequently at higher temperatures (25 and 30°C).

Although some workers find similar distribution patterns using selective media and insect baiting methods (Keller, Kessler, and Schweizer 2003) others such as Bruck (2004) noted a disparity between selective media and insect baiting in fungal recovery. It is generally accepted that isolation-based approaches give a biased view of many microbial systems.

Molecular methods

To overcome the disadvantages of the isolation-based approaches, much attention has focused on molecular methods which include restriction fragment length polymorphism, random amplified polymorphic DNA polymerase chain reaction (RAPD-PCR), microsatellite markers and strain specific probes (Enkerli, Widmer, Gessler, and Keller 2001; Pantou and Typas 2003, Enkerli, Widmer, and Keller 2004; Wang, Fan, and Butt 2004; Dolci, Guglielmo, Secchi, and Ozino 2006; Enkerli, Ghormade, Oulevey, and Widmer 2009; Oulevey, Widmer, Kölliker, and Enkerli 2009). DNA-based characterisation techniques have the advantage that specific genes can be amplified from a community mixture or pure culture by PCR and that products of such amplifications can be further characterised e.g., by subcloning and DNA sequencing (Ranjard, Poly, Lata, Mougel, Thioulouse, and Nazaret 2001). Such data can be directly compared to DNA sequence databases, providing information about similarity to already known genes. Quantitative (real time) PCR methods have been developed to quantify individual species and in some cases strains of a specific species of fungus (Atkins, Clark, Pande, Hirsch, and Kerry 2005; Rubio, Hermosa, Keck, and Monte 2005; Cordier, Edel-Hermann, Martin-Laurent, Blal, Steinberg, and Alabouvette 2007; Castrillo, Griggs, and Vandenberg 2008; Shah and Butt, unpublished results). Besides revealing more about the putative natural levels of EPF, molecular tools have also been used to study a range of other parameters which would be of interest to regulatory authorities or risk assessors such as parasexual recombination and the impact of exotic strains on indigenous populations (e.g., Leal-Bertioli, Peberdy, Bertioli, and Butt 2000; Wang et al. 2002, 2004). Molecular methods have many benefits but also have limitations as reviewed by Juste, Thomma, and Lievens (2008), Kent and Triplett (Kent and Triplett 2002) and Kirk et al. (2004). The most important limitations are that molecular methods may not distinguish between healthy and senescent cells nor provide much information on the virulence or host range of the EPF. It should be noted that highly sensitive molecular techniques have only been developed in the last decade. Previously less sensitive methods were deployed such as isoenzyme polymorphism (Rakotonirainy, Cariou, Brygoo, and Riba 1994; Bridge et al. 1997) to distinguish between natural and/or introduced populations of EPF.

In this review, data were collected from the literature in which concentrations of EPF were expressed in CFU/g soil or could be converted into CFU/g soil from other similar units. Effectively, only data deriving from studies using selective media could be used.

Natural background concentrations

This section reviews the natural levels of fungi recorded in the soil of diverse wild and managed habitats (i.e., before application of the EPF) and briefly explains the data presented. The attempt was made to include all published work. It does not cover natural populations of endophytic EPF as this has been reviewed extensively by Vega and co-workers (Vega 2008; Vega et al. 2008, 2009). Figures 1–3 summarise the natural levels reported for three commonly studied EPF, M. anisopliae, B. bassiana and B. brongniartii. All densities were determined in samples collected from the field, sometimes during expanded periods of time. Detailed information on the sources of the natural background levels is given in Table 1. All data derived from studies in which selective media were used.

Figure 1.  Background concentrations of M. anisopliae.

Figure 2.  Background concentrations of B. bassiana.

Figure 3.  Background concentrations of B. brongniartii.

Table 1. Natural concentrations of (a) Metarhizium anisopliae; (b) Beauveria bassiana; and (c) Beauveria brongniartii.

Study no.	Soil type	Crop	CFU/g soil	References	Remarks
(a) Metarhizium anisopliae
1a	n.r.	Meadow, orchard, arable land	350–4500	Keller et al. ( 2003)	Hosts present. Switzerland
1b	n.r.	Meadow, orchard, arable land	115–6800	Keller et al. ( 2003)	Hosts absent. Switzerland
2	n.r.	Lucerne field	0–300	Vestergaard and Eilenberg (2000)	Denmark
3a	Sand	Cabbage, oil seed rape, cereals	0–4600	Vänninen et al. (2000)	Cylinders buried in the field. Finnish isolate. Northern Finland
3b	Clay	“	0–400
3c	Peat	“	0–1000
4a	n.r.	Soybean, wheat, oat	0–42,500	Sosa-Gomez et al. (2001)	Till conditions. Brazil
4b			0–3200		No-till conditions. Brazil
5a	Oxisol^1	Soybean, wheat	70–1030	Sosa-Gomez and Moscardi (1994)	Till conditions. Brazil
5b			90–12,650		No-till conditions. Brazil
6	Silt loam	Pasture	4300 (average)	Rath et al. (1995b)	Hosts present. Strain DAT F-001. Tasmania
7	Sandy clay loam	Pasture	0–130,000	Rath and Bullard (1997)	Presence of hosts not confirmed. Strain DAT F-001. Tasmania
8	Silt loam, sandy clay loam, peat, loam sandy loam and loamy sand		<100 for all isolates	Roddam and Rath (1997)	No correlation between soil type and abundance strain DAT F-494-499 Subantartic Macquarie island
9	n.r.	Pasture	0	Rath et al. (1995a)	Tasmania
10a	n.r.	Pasture	Until 10^6	Milner (1992)	Australia
10b			Until 2000		Australia
11	Red volcanic, sandy silt loam, sandy silt loam, sandy loam, loam, clay	All sugar cane sites	3000 (average) 8000 (average) 3000 (average) 3000 (average) 15,000 (average) 10,000 (average)	Milner et al. (2003)	Australia
12	n.r.	Golf course	2000 (average) 600 (average)	Strasser et al. (2005)	In presence of heavy infestation of Phyllopertha horticola (497 larvae/m^2). Austria
13	n.r.	Paddock	Up to 3000	Townsend et al. (1995)	Infestation with grass grub Costelytra zealandica. Matangi, New Zealand
(b) Beauveria bassiana
1	Oxisols^1	Soybean, wheat	0–190,000	Sosa-Gomez and Moscardi (1994)	2 years; Dec-April; till and no-till systems; data by approximation from graph; 16 sampling data below 500 CFU/g soil
2	Sandy loam	Grassland	<10–16,000	Gaugler et al. (1989)	Sampling during one year
3	n.r.	soybean, wheat, oat	0–29,400	Sosa-Gomez et al. (2001)	Nov-March; till and no-till systems; data by approximation from graph; 13 sampling data below 10 CFU/g soil, 6 sampling data between 500 and 1000 CFU/g soil.
4	n.r.	Corn	51–74 (averages)	Bing and Lewis (1993)	Mean numbers in plow, chisel and no-tillage systems; SD or SE not reported
5	n.r.	Paddock	1000 to 28,000	Townsend et al. (1995)	Matangi, New Zealand, infestation with grass grub Costelytra zealandica
6	Clay-loam	Alfalfa wheatgrass	9–40 (averages)	Inglis et al. (1997)	No differences in numbers of CFU in alfalfa and wheatgrass. Possibly contamination in control plots as a result from drift from treated plots
7	Silt loam, sandy clay loam, peat, loam sandy loam and loamy sand		<100	Roddam and Rath (1997)	Strain F140 Subantartic Macquarie island
8	Loam	Paddock	742–5785	Brownbridge et al. (2006)	Biopolymer-coated rice. Sampled at 0, 3, 6 and 12 weeks. High uniform uniform population of clover root weevil
9	Brown forest soil	Pinus forest	0–1600	Shimazu et al. (2002a)	Monthly samplings during 456 days, Japan
10a	Forest soil	Pinus thunbergii forest	0–400	Shimazu et al. (2002b)	Monthly sampling during 3 years. Japan (Futtsu, Chiba)
10b	Forest soil	Quercus forest	0–470,000	Shimazu et al. (2002b)	Monthly sampling during 3 years. Japan (Kukisaki, Ibaraki)
10c	Forest soil	Chamaecyparis Forest	0–35,000	Shimazu et al. (2002b)	Monthly sampling during 3 years. Japan (Kukisaki, Ibaraki)
11	Not specified	Farmers's field	13,000	Devotto et al. (2007)	Chile
(c) Beauveria brongniartii
1a	Silty and clay loam	Meadows	0–300	Kessler et al. (2004)	Melolontha melolontha. Absent
1b	Sandy and clayloam		0–4300		Melolontha melolontha present. Switzerland
2	n.r.	Meadows	0–808	Kessler et al. (2003)	Presence of hosts not documented. Strain FAL 546, Switzerland
3	n.r.	Meadows	80–1060	Keller et al. (2003)	Only in Presence of Melolontha melolontha. Strain FAL 546, Switzerland
4	n.r.		≤10^3	Cravanzola et al. (1996)	Italy
5	Sandy soil	Potato	130–3700	Laengle et al. (2005)	A mean of 40 grubs of Melolontha melolontha Austria
6	Various soils	2 orchards 6 meadows	0–33 0–808	BIPESCO consortium	Switzerland
^1Soil taxonomy of Brazil = latosolo roxo distrófico epieutrófico.
n.r., not reported.

The data are presented in Box and Whiskers plots with boxes extending from the 25th to the 75th percentile. The lines in the middle of the boxes represent the medians. A 95th percentile of the geometric mean of the number of colony forming units (CFUs) was calculated for each species. Firstly, suitable studies were selected. Data from one study were not used when only the highest and/or the lowest value was given or when only the mean value was given. An exception was made when means were derived from several replicates. Secondly, for each suitable study a log geometric mean and its upper 95th percentile was calculated. Finally, the overall geometric mean was calculated for the selected studies as the average of the individual log observations, while accounting for the between study variation. The overall geometric mean was then the exponent of this value. The derived 95th percentile of the geometric mean was choosen to represents the upper natural background level. By choosing the 95th percentile some very high peaks were excluded. These high peaks may be due to errors in determining CFU numbers.

In Figure 1 natural background concentrations are given for the entomopathogenic species M. anisopliae. The 95th percentile of the geometric mean was 1040 CFU/g soil based on studies 1a and 1b, 2, 3a–c, 4a and 4b, 5a and 5b, 7 and 11. The data clearly show that M. anisopliae is a cosmopolitan species as studies were performed in Europe (Switzerland, Denmark, Austria, Finland), South America (Brazil), Australia (including Tasmania), New Zealand and the Macquarie island. A similar picture was given by Zimmermann (2007b) who also collected data on natural occurrence of this fungus. Many of the data collected by this author could not be used in Figure 1 as they were obtained with the Galleria baiting technique.

In Figure 2, natural background concentrations are given for the entomopathogenic species B. bassiana. The 95th percentile of the geometric mean was 830 CFU/g soil based on studies 1–3 and 8–10. Unfortunately, the authors of the studies did not document data on the existing host population. Without this information, it is not possible to relate extremes in numbers of CFU to the presence of high host populations.

In Figure 3, natural background concentrations are given for B. brongniartii. The 95th percentile of the geometric mean was 740 CFU/g soil based on studies 1a, 1b, 2, 3, 5 and 6. Studies by Keller et al. (2003) and Kessler, Matzke, and Keller (2004) suggest that B. brongniartii is particularly prevalent in areas were its host, M. melolontha, is present (Table 1).

Levels of introduced EPF

Figures 4–10 summarise the fungal densities reported for various laboratory (small scale) and field studies (pastures, orchards and arable fields). Where comparable experiments were performed within one article with similar outcomes, we selected one or two representative experiments. Studies by virtually all the different workers report a decline in spore levels over time. Factors contributing to this decline are covered below in the section ‘Factors explaining the decline in inoculum levels’. Detailed information on the inoculum levels, habitats and source of the data are given in Tables 2–5.

Figure 4.  Metarhizium anisopliae laboratory and small scale experiments.

Figure 5.  Metarhizium anisopliae field experiments in pastures.

Figure 6.  Metarhizium anisopliae field experiments in arable crops.

Figure 7.  Beauveria bassiana laboratory and small scale experiments.

Figure 8.  Field experiments with Beauveria bassiana.

Figure 9.  Beauveria brongniartii experiments in pasture and orchards.

Figure 10.  Lecanicillium longisporum laboratory experiments.

Table 2. Persistence of Metarhizium anisopliae in the soil.

Strain	Soil type	Crop	Initial dose/g soil	References	Remarks
Field experiments
Wild type strain ARSEF1080 recombinant gpd-Pr1-4	Sandy loam	Cabbages	10^13/ha 1.3×10^4/g soil	Hu and St. Leger ( 2002)
Strain 172		Irrigated rice field	7.2×10^13 CFU/ha	Martins et al. (2004)	Brazil
DAT F-001	Sandy clay loam	Pasture	5.1×10^4	Rath and Bullard (1997)	Tasmania, initial rise coincided with presence L3 larvae and pupae
DAT F-001	Silt loam	Pasture	5.1×10^4	Rath et al. (1995b)	Tasmania
DAT F-001	Sandy clay loam	Pasture	6.6×10^4	Rath et al. (1995a)	Tasmania, a host population was present
			3.3×10^4
			6.3×10^4
var. lepidiotum FL-147	Loam clay sandy loam	Sugar cane	10^5	Milner et al. (2003)	Australia; rice based granules^1; from original data obtained from Milner
var. anisopliae FL-1045	Sandy silt loam sandy silt loam red volcanic	Sugar cane	10^5	Milner et al. (2003)	Australia; rice based granules^1; from original data obtained from Milner
Strain 275 (ATCC 90448)		Lucerne field	10^9/m^2 10^9/m^2 10^11/m^2 10^11/m^2	Vestergaard and Eilenberg (2000)	Denmark; hosts were present; two replicates
Bipesco 6 Commercial strain Schweizer Commercial strain Andermatt Spore powder Bipesco 5		Golf course	50 kg granules/ha	Strasser et al. (2005)	In presence of heavy infestation of Phyllopertha horticola (497 larvae/m^2)
BIO1020	Clay peat	Nursery stock Thuja occidentalis	1 g granules/L soil	Van Tol (1993)	The Netherlands; field small scale
BIPESCO 5	Not documented	Abies procera plantation	1.0×10^14 condia/ha 0.4 − 1.0×10^14 condia/ha	Nielsen et al. (2004)	Denmark
Laboratory experiments
Strain F52	Peat (commercial mixture for container growth)		1.9×10^6	Bruck (2005)	Pots in greenhouse; hosts present; >95% infection
BIO1020	Peat/sphagnum (commercial mixture for container growth)	Nursery stock Thuja occidentalis	1 g granules/L soil	Van Tol (1993)	The Netherlands; pots
SF85-60	Peat sand clay		10^8/cm^2 2×10^7/cm^3 soil^1	Vänninen et al. (2000)	Northern Finland; soil filled cylinders in field; data from upper 5 cm
SF86-51	Sand clay peat		10^8/cm^2 2×10^7/cm^3 soil^1	Vänninen et al. (2000)	Finland
Strain no. 51	Sandy loam, 80% moisture holding capacity		8×10^9	Fargues and Robert (1985)
Strain no. 32	Sandy loam, 80% moisture holding capacity		12×10^9 CFU/g dry fungal material	Fargues and Robert (1985)
^1Assuming distribution of the spores in the upper 5 cm of the soil; percentage of recovery given in the table also from the upper 5 cm.
^2In colonies per g soil. A correction factor of 10^4 has been applied to the data given in the publication after personal contact with the author. This was the dilution factor which had not integrated into the data yet.

Table 3. Persistence of Beauveria bassiana in the soil.

Strain	Soil type	Crop	Initial dose	Method	References	Remarks
ARSEF-252	Sandy loam	Grassland	5×10^6 CFU/cm^2 soil^2	Oatmeal-dodine agar	Gaugler et al. (1989)	United States; autumn tilted appl.
ABG 6112	Sandy clay loam	Soybean/sorgum	9.4×10^4 CFU/cm^2	Sabouraud's dextrose medium	Storey et al. (1989)	United States; field; tillage and no tillage, resp.; no hosts mentioned; upper 5 cm
No. 18	Sandy loam		20×10^10 CFU/g soil	Medium with chloramphenicol	Fargues and Robert (1985)
SF86-23	Sand peat clay		10^8 CFU/cm^2 2×10^7 CFU/cm^3 soil^1	Modified OAES medium with chloramphenicol and cycloheximide	Vänninen et al. (2000)	Finland; presence of hosts not mentioned
Firma Nutrilite products	Sandy loam		1×10^7 CFU/g soil	Semi-selective medium	Müller-Kögler and Zimmermann (1986)	Laboratory experiment
RS-252	Silt soil	Potatoes	2.8×10^10 CFU/m^2 equal to 3.7×10^5/g soil	Heck's selective medium	Watt and LeBrun (1984)	Kingston, Rhode Island, USA Detection limit of 7.4×10^3 propagules/g soil
ABG 6112	Sandy loam		1.7×10^12 CFU/hectare equal to 2.3×10^3 CFU/g soil	Oatmeal-dodine agar	Storey et al. (1987)	Tifton, GA, USA
F418	Loam	Paddock	2×10^13 CFU/hectare equal to 2.7×10^4 CFU/g soil	PDA with streptomycin sulphate, tetracyline hydrochloride and cycloheximide	Studdert and Kaya (1990a)	Biopolymer-coated rice, 4 samples in 5 replicates during 12 weeks. High uniform uniform population of clover root weevil
F-263	Brown forest soil	Pinus forest	1.4×10^12 CFU/m^2 (mixed to depth 30 cm) Equal to 3.1×10^9 CFU/g soil	D0C2	Shimazu et al. (2002b)	Japan, monthly samplings during 456 days
GHA, Mycotech Corp.	Clay-loam	Wheatgrass	5×10^13 CFU/ha equal to 6.7×10^6 CFU/g	Semi-selective oatmeal-dodine medium	Inglis et al. (1997)	Alberta, Canada
Strain QU-B931 collected from Dalaca pallens	Not defined	Hybrid ryegrass	10^12 CFU/ha equal to 1.3×10^5 CFU/g soil	Selective medium	Devotto et al. (2007)	Osorno, Chile
^1Assuming distribution of the spores in the upper 5 cm of the soil; percentage of recovery given in the table also from the upper 5 cm.
^2Soil tillage to a depth of 7.5 cm.

Table 4. Persistence of Beauveria brongniartii in the soil.

Strain	Soil type	Crop	Initial dose (kg/ha)	References	Remarks
Local isolates		Apple orchard		Dolci et al. (2006)	Hosts present
		Meadow	Unknown	Keller et al. (2003)	Hosts present
		Meadow orchard	40–50	Kessler et al. (2004)	Hosts absent
		Meadow Orchard	40–50	Kessler et al. (2004)	Hosts present
Local isolate/grown on barley		Grassland	Unknown	Enkerli et al. (2004)	Hosts present in field but numbers
Local isolate/ grown on barley		Apple			not specified
Local isolate/ blastospore		Orchard
MELOCONT	Grey warp soil Pararendzina	Pasture Pasture	20–39 20–39	Strasser (1999,2001)	>20 grubs of Melolontha melolontha in the first year of the trial
	Pararendzina	Pasture	20–39		5 application data

Table 5. Persistence of Lecanicillium longisporum (formerly Verticilium lecanii) in the soil.

Species	Soil type	Initial dose/g soil	No. of sampling dates	Max CFU/g soil	CFU/g soil at end	Duration relevant sampling period	Method	References	Remarks
Lecanicillium longisporum, strain V24	Garden soil	1.9×10^7	7	1.3×10^7	4×10^6 2×10^4 10^4	40 days	Selective agar plates	Beyer et al. (1997)	Laboratory experiments in pots; at 5, 20 and 30% water content; at 20°C
Lecanicillium longisporum, strain V24	Garden soil	1.9×10^7	7	10^7	1.3×10^6 5×10^5 316	40 days	Selective agar plates	Beyer et al. (1997)	Laboratory experiments in pots; at 20, 25 and 30°C and 30% soil moisture

Pattern of decline noted for M. anisopliae

Fungal densities of M. anisopliae collected from applications in laboratory or (small scale) field studies are graphically presented in Figures 4–6.

It is clear that the decline of M. anisopliae conidia is quite similar in field experiments (pasture and arable fields) (Figures 5 and 6). Persistence studies in pastures were mostly done over a period of 1 year except for the studies of Rath, Koen, and Anderson (1995a), Rath et al. (1995b) and Rath and Bullard (1997) which were conducted over a period of 2.5, 3 and 7.5 years (Figure 5). An initial increase to 5×10⁵ conidia/g soil in month 12 of the studies of Rath et al. (1995b) and Rath and Bullard (1997) coincided with the occurrence of many larvae of the target pest, Adoryphorus couloni. These larvae were found dead and sporulating, explaining the increase. Thereafter, a decline was noted, initially fast and becoming slower later on. At the end of these experiments, M. anisopliae conidia levels remained approximately a factor 100 higher than the upper natural background level of 1040 CFU/g soil even after 7.5 years. Also in arable fields (Figure 6), fungal spore numbers in the study of Milner, Sampson, and Morton (2003) remained up to a factor 300–1000 higher than the natural background level after 42 months (3.5 years). They, however, showed a slow but steady decline. Other studies also noted a decline of conidia levels with the exception of Strasser, Zelger, Perfuss, Längle, and Seger (2005) who noted an increase in fungal density of M. anisopliae BIPESCO 5 after 20 months. This increase is most likely explained by heavy infestation of the target pest, Phyllopertha horticola (497 larvae/m² and with a maximum of 1574 grubs per m²). It is evident that in this situation the recycling is greater than the decrease of the initial inoculum.

Fungal density can also be increased by multiple applications. In the study of Nielsen, Eilenberg, Harding, and Vestergaard (2004) two extra applications were performed 10 and 11 months after test initiation (Figure 6). Unfortunately, samples were not taken in month 11, at which time the largest increase of inoculum would be expected. The effects were still visible in month 14 as the fungal density was slightly elevated as compared with the fungal density in month 6. Vestergaard and Eilenberg (2000) showed very similar degradation rates of M. anisopliae in both the high and low density experiments done in arable systems (Figure 6). A similar degradation rate was found in the experiments of Hu and St. Leger (2002). A much faster degradation was found in flooded rice fields (Martins, Botton, Carbonari, and Quintela 2004) which supports the findings of others that spore survival is poor in wet soils (see section ‘Factors explaining the explaining the decline in inoculum levels’).

The pattern of decline in laboratory and small scale experiments is for most experiments similar to the field experiments even though inoculum was applied at much higher concentrations (Figure 4). Fargues and Robert (1985) noted that one M. anisopliae strain (no. 32) declined rapidly whereas another strain (no. 51) declined more slowly demonstrating differences in persistence or ecological fitness. It should be noted that strain 51 had a degradation rate lower to that reported by other workers (Figure 4).

Most studies in Figures 4–6 did not show a decline to the upper natural background level of 1040 CFU/g soil within the time span of the experiments. Apart from temporary increases due to reproduction in the host, fungal numbers show a steady decline in all studies. The upper natural background level is reached almost 10 years post-treatment. This is however a rough estimation as most studies did not last long enough to reach the background level.

Pattern of decline noted for Beauveria bassiana

Figure 7 (lab and small scale experiments) and 8 (field experiments) summarise the persistence reported for B. bassiana by various workers. It shows a decline in conidial levels under different agronomic conditions. Some workers report a rapid decline in levels of introduced B. bassiana. For example, Müller-Kögler and Zimmerman (1986) mixed B. bassiana conidia in soil at a rate of 10⁶ to 10⁷ conidia/g soil dw, and found that the number of conidia was reduced by 1/100 to 1/1,000 one year later. Shimazu, Maehara, and Sato (2002a) also noted a gradual decline of conidia of B. bassiana mixed in pine forest soils to 1/10 of the initial density 12 months after application. Vänninen et al. (2000) noted differences in degradation in clay, sand and peat with survival being least in peat.

Field studies conducted by Brownbridge et al. (2006) using B. bassiana to control clover weevil, typically show the variability in the results (Figure 8). At some sites more inoculum was recovered from control than treated plots and the rate of decline was greater at some sites than others. In fact, there was considerable variation in the occurrence of conidia making it difficult to statistically differentiate between fungal BCA treated and untreated control sites 12 weeks post treatment. However, at some sites a classical pattern was evident, i.e., temporary elevation of inoculum immediately after application then a decline in levels over time.

Devotto, Cisternas, Gerding, and Carrillo (2007) reported the natural level of B. bassiana spores in grassland soil was 1.3×10⁴ CFU/g of dry soil which increased 70% after spraying, but then declined dramatically with the second post-spraying estimate not differing from pre-treatment levels (Figure 8).

In Figures 7 and 8 not all experiments lasted long enough to show that the fungal concentrations decline to the upper natural background level of 830 CFU/g soil. The overall picture however shows that there is a consistent decline of fungal concentrations. The upper natural background level is approximately reached after 0.5–1.5 years.

Pattern of decline noted for Beauveria brongniartii

Figure 9 summarises the persistence of B. brongniartii in different agronomic systems. Inoculum of this fungus, as in the case of B. bassiana and M. anisopliae, gradually decrease in a very similar way. Some strains of B. brongniartii have been shown to persist albeit at low densities (<10³ CFU/g soil dwt) 7–14 years after application (Keller et al. 2003; Enkerli et al. 2004). Kessler et al. (2004) showed that survival of B. brongniartii 16 months post application in host and host-free sites was 78 and 12% of the initial inoculum, respectively. These observations suggest that B. brongniartii cannot survive for long in the absence of its host. In a trial of Strasser (1999), and additional data of final sampling data of this study in (Strasser and Enkerli 2001), initial increases of B. brongniartii were demonstrated at a host population of only >20 grubs of the cockchafer Melolontha melolontha per m², but in this case the increase was probably caused by the repeated applications (five in 3 years). By performing these repeated applications it was shown that it is possible to compensate for factors such as suboptimal soil conditions and too few grubs to proliferate on. These elevated levels, however, dropped immediately once the applications had stopped.

In Figure 9, not all experiments lasted long enough to show that the fungal concentrations decline to the natural background levels of 740 CFU/g soil. The overall picture however shows that the upper natural background level is reached after approximately 4 years.

Pattern of decline noted for Lecanicillium longisporum

Only one laboratory experiment with Lecanicillium longisporum could be obtained from the literature (Beyer, Hirte, and Sermann 1997). In this experiment, conidia survived best at 5% moisture and 20°C with less than a factor 10 (Figure 10). Persistence decreased at higher moisture rates and temperatures with factors 100 to >100,000.

Conclusions on decline of inoculum

In these studies, samples were mainly taken from arable lands and meadows in which the inoculum was also to be applied. The upper natural background concentration is represented by the 95th percentile of the geomean of the number of colony forming units (CFUs) obtained from different studies. The upper natural background levels were 1040 CFU/g soil for M. anisopliae, 830 CFU/g soil for B. bassiana and 740 CFU/g soil for B. brongniartii. These numbers are to be treated as estimations and a general number of 1000 CFU/g soil can be employed for all three species.

The data all show a decline of fungal persistence. It was estimated that the applied inoculum density decreases to upper natural background levels within 0.5–1.5 years for B. bassiana, after about 4 years for B. brongniartii and >10 years for M. anisopliae.

Factors explaining the decline in inoculum levels

We have shown that EPF are widespread and that natural inoculum levels in the soil vary with habitat. Furthermore, we have shown that introduced inoculum will decline with time. Researchers have endeavoured to improve the performance of EPF in a variety of ways such as improving formulations, genetic transformation, combining with other agents (e.g., synthetic chemicals, entomopathogenic nematodes, neem) to obtain synergistic effects (Butt et al. 2001; Fan et al. 2007; Shah, Ansari, Prasad, and Butt 2007a; Wang and St. Leger 2007; Ansari, Shah, and Butt 2008; Shah, Gaffney, Ansari, Prasad, and Butt 2008). So what are the factors regulating inoculum levels? A wide range of factors contribute to their decline. These factors can be broadly divided into the following categories: intrinsic, edaphic, biotic, cultural and climatic.

Intrinsic attributes

The intrinsic stability of fungal propagules (submerged conidia, aerial conidia and blastospores) depends on the strain and cultural conditions. The latter include culture media which could influence conidial endogenous reserves and water activity. Harvesting-drying methods and storage conditions will also influence shelf life. Some of these intrinsic attributes are strongly interlinked. They are briefly commented on below.

EPF strains

EPF strains differ in their ‘ecological fitness’ i.e., ability to persist in the field and successfully infect a host under sub-optimal conditions (Butt 2002). The underlying mechanisms to explain this resiliance has not been elucidated but may explain why certain strains are found in certain habitats or have greater tolerance of the environmental constraints outlined below. There is much evidence for these intrinsic genetic differences. For example, certain genotypes of M. anisopliae and B. bassiana more tolerant of UV and high temperatures were reported to be predominant in soils of Canadian agro-ecosystems over isolates from cooler, shaded forested habitats of similar latitude (Bidochka, Kamp, Lavender, Dekoning, and De Croos 2001). Meyling, Lübeck, Buckley, Eilenberg, and Rehner (2009) investigated species diversity, reproductive potential and genetic structure of Beauveria occurring within a single arable field and bordering hedgerow in Denmark. Seven species were present throughout the hedgerow habitat but significantly, only B. bassiana s.s. phylogenetic species Eu_1 was isolated from tilled soils. Often these intrinsic differences between strains are masked by the complex environmental and climatic factors impacting on the inocula, some of which have been explored in this article and also reviewed by Inglis, Goettel, Butt, and Strasser (2001) and Klingen and Haukeland (2006).

Cultural conditions

These can influence the form of the propagule. Solid media usually give rise to aerial conidia while liquid media can give rise to thin-walled blastospores. Aerial conidia are generally more robust than the blastospores (Humphreys, Matewele, Trinci, and Gillespie 1989; Hegedus, Bidochka, and Khachatourians 1990). Manipulation of the culture media (osmolarity, C:N ratio) can improve virulence (Ibrahim, Butt, and Jenkinson 2002; Leland, Mullins, Vaughan, and Warren 2005a, 2005b; Shah, Wang, and Butt 2005; Safavi et al. 2007) and influence the robustness of conidia during drying. Medium components and culture conditions affect physiological parameters which subsequently influence the shelf life or ecological fitness of the inoculum. For example, cultural conditions can influence the thermotolerance of aerial conidia of B. bassiana (Ying and Feng 2006) or ecological fitness of M. anisopliae blastospores (Ypsilos and Magan 2005).

Harvesting-drying of inoculum

The drying process (e.g., lyophilisation, spray or fluid bed drying) has probably the most profound impact on viability/shelf life (Horaczek and Viernstein 2004). Conidia produced on osmotic stress media are relatively more stable to drying than blastospores or submerged conidia. Jackson and Payne (2007) showed that the relative humidity (RH) of the drying air significantly affects the desiccation tolerance and the storage stability of P. fumosoroseus blastospores. The degree to which conidia are dried can influence their susceptibility to imbibitional damage (Faria, Hajek, and Wraight 2009).

Storage of inoculum

Freshly harvested conidia usually quickly lose viability when stored at room temperature (Moore, Douro-Kpindou, Jenkins, and Lomer 1996; Butt et al. 2001). The germination rate of dry conidia of M. flavoviride stored as a powder at 10–14°C is ca. 95% germination but can be less than 27% if stored at 28–32°C (Moore et al. 1996). Over 90% germination was observed for dry conidia stored over 128 days with or without silica gel at temperatures ranging between 10 and 18°C (Moore et al. 1996). Similar patterns are observed for other EPF. For example, the viability of dry conidia of B. bassiana was 635 days at 8°C and 0% RH in contrast to 28–56 days at 25°C and 75.8% RH (Zimmermann 2007a). In other words, dry conidia survive longer in soils at low humidities and temperatures. According to Chen, Shi, and Zhang (2008) temperature was the most critical factor influencing conidial storage stability of V. lecanii. Viability decreased with increasing storage temperatures with virtually no spores surviving storage at 35°C for 6 months. Chen et al. (2008) considered the moisture content of V. lecanii conidia as another major factor influencing their viability when stored at room temperature; conidia dried to ca. 5% moisture content showed higher viability than non-dried conidial powder. Conidia of M. anisopliae have a longer shelf life if incorporated into soil-less media (e.g., peat) than if stored without any carrier at ambient temperature and humidity (Butt unpublished observations). In the absence of suitable nutrient substrates, fungal propagules will ultimately exhaust their endogenous reserves and die.

Edaphic

Soil texture, organic matter content, pH and water potential can have a direct or indirect effect on the persistence of conidia of insect-pathogenic fungi. The occurrence and persistence of EPF may be inter-linked: it seems logical to assume that if a soil is not conducive for an entomopathogenic fungus, then it will not be found at that site or will persist poorly if used at that site. Based on this assumption, this section will touch on the occurrence as well as survival (persistence) of inocula in different soil types.

Soil texture

Soil texture can influence percolation of inoculum through the soil profile and spore acquisition by insects. If the fungal inoculum leaches to a depth beyond that inhabited by the host, infection will not occur and there is little chance of inoculum levels increasing. Sandy soils and soils low in organic matter generally retain fewer propagules than clay-textured and organic soils (Ignoffo, Garcia, Hostetter, and Pinnel 1977; Storey and Gardner 1987, 1988). Vänninen et al. (2000) made similar observations when studying the persistence and penetration into soil of surface-applied unformulated conidia of two isolates of M. anisopliae and one of B. bassiana. These workers found that M. anisopliae was the most persistent with clay being the most, and peat the least favourable soil for persistence and as expected conidia penetrated peat more than clay (Figures 6 and 7). Vänninen et al. (2000) also noted differences in persistence between the two M. anisopliae isolates in the sandy and peat sites, but not in the clay site. In contrast, B. bassiana disappeared during the first winter after application to clay and to one of the two sandy sites but was detected 1 year later in the peat and the other sandy site. Recent studies show that M. anisopliae can survive for several months in soil-less horticultural growing media which include peat, coir and bark (Bruck and Donahue 2007, Butt, Shah and Ansari, unpublished observations). Furthermore, conidia of M. anisopliae percolate more readily through the soil-less media if applied as a drench than if premixed into composts (Shah, Prasad, and Butt 2007b).

Recently, Quesada-Moraga, Navas-Cortes, Maranhao, Ortiz-Urquiza, and Santiago-Álvarez (2007) found that irrespective of habitat type, B. bassiana predominated over M. anisopliae in soils with a higher clay content, higher pH, and lower organic matter content. Logistic regression analyses showed that pH and clay content were predictive variables for the occurrence of B. bassiana, whereas organic matter content was the predictive variable for M. anisopliae. These workers established that, irrespective of geographical location, absence of both fungal species was determined by alkaline sandy soils with low organic matter content, whereas heaviness of soil texture, acidity and increasing organic matter content led to progressively higher percentages of samples containing EPF. Similar observations were recorded in field trials done in New Zealand where Brownbridge et al. (2006) applied a polymer coated B. bassiana for clover weevil control and found elevated inoculum levels up to 6 weeks post-treatment in the Waikato loamy soils, but a more rapid decline in the sandy Manawatu soils.

Whereas some workers suggest that edaphic factors may influence the occurrence and persistence of EPF, others appear to find these organisms in a range of soil types. For example, Asensio, Carbonell, Lopez-Jimenez, and Lopez Llorca (2003) isolated B. bassiana, M. anisopliae and L. lecanii from a range of soil textures in Alicante, Southern Spain, including sandy loam, clay, loam, loamy sand, loamy-clay-silt. More EPF were recovered from forest soils than agricultural (orchard) soils (62.5 vs. 50%) independent of soil texture (Asensio et al. 2003). Bidochka, Kasperski, and Wild (1998) isolated EPF (B. bassiana, M. anisopliae, Paecilomyces species) from 91% of the sites tested across Ontario, Canada and concluded that the occurrence of M. anisopliae and B. bassiana was not related to soil texture or pH.

Soil texture also affects the survival of the nematophagous fungus Paecilomyces lilacinus. For example, Rumbos, Mendoza, Sikora, and Kiewnick (2008) noted a gradual decline in the densities of P. lilacinus over time. The application rate had no significant effect on the population dynamics of this fungus and its decline was not significantly affected by the presence of the nematode host. However, soil texture did have a significant effect on its persistence; P. lilacinus persisted longer in silty loam and clay soil, with reduced persistence when sand was added to field soil. Persistence increased significantly when an organic substrate was added to pure sand (Rumbos et al. 2008).

Soil moisture

Soil moisture (water potentials) influences the persistence of some EPF with few conidia surviving wet soils (Inglis et al. 2001). Li and Holdom (1993) found that survival of M. anisopliae conidia 160 days after application was 20 and 40% in wet and dry soils, respectively.

Higher soil moisture has a negative effect on survival of EPF such as Lecanicillium longisporum (Beyer et al. 1997, Figure 10). Studdert and Kaya (1990b) mixed conidia of B. bassiana in non-sterile fine sandy loam or peat soil at water potentials ranging from 0 bars (saturation) to −1500 bars and found that conidia half lives were longest at −15 bars and decreased as the water potential approached either 0 or −200 bars. Conidia half lives increased as the water potential decreased from −200 to −1500 bars (i.e., soils became drier). These workers established that the longest mean half life value for B. bassiana was 44.4 weeks for conidia incubated in the fine sandy loam soil at −10 bars and 10°C and the shortest half life value was 0.3 weeks in peat soil at 0 bars and 28°C.

Soil pH

The influence of soil pH on the persistence and efficacy of EPF has not been thoroughly investigated. The fact that EPF have been isolated from a wide range of soils and habitats suggest that pH is not an impediment. Jaffee and Zasoski (2001) found that the nematophagous fungus Hirsutella rhossiliensis was more active (i.e., parasitized more assay nematodes) in an acidic vineyard soil than in a neutral vineyard soil. Heating the neutral soil to 60°C for 2 h did not alter soil pH or electrical conductivity but increased fungus activity to levels equivalent to those in acidified soil, suggesting that the soil biota may have been influencing the efficacy of H. rhosiliensis in the neutral soil.

Biotic

Soils are complex environments with a rich, diverse composition of macro and micro-organisms. Fungal BCAs are subject to parasitism, predation and antagonism by a wide array of soil organisms including other fungi (mycoparasites), bacteria, protozoa, mites, nematodes, collembolans, and annelids (Maraun, Martens, Migge, Theenhaus, and Scheu 2003). Table 6 lists examples of mycophagous organisms. Micro-organisms are implicated as the causal agents for fungistasis, with their action mediated either by nutrient deprivation or production of antifungal compounds. Although the impact of these consumers and antagonists on the persistence of EPF has not been quantified, it is generally accepted that they could reduce levels of indigenous and introduced fungi (Rosenheim 1998; Inglis et al. 2001; Hasna, Insunza, Lagerlof, and Ramert 2007). The persistence of EPF is significantly reduced in non-sterile vs. sterile soils (Lingg and Donaldson 1981; Pereira, Stimac, and Alves 1993) clearly demonstrating the suppressive role of soil biota. However, soil sterilization studies do not differentiate between the contribution of predation, antibiosis, and competition on the persistence of fungal BCAs.

Table 6. Mycophagous soil organisms.

Mycophagous organism	Example	Reference
Microfauna
Nematodes	Aphelenchoides, Aphelenchus, Tylenchus, Ditylenchus	Bae and Knudsen (2001); ,1985); Giannakis and Sanders (1989); Hasna et al. (2007); Ruess et al. (2000)
Protozoa	Amoebae (e.g., Phryganella acropodia, Ripidomyxa australiensis, Thecamoeba granifera, Trichamoeba mycophaga) Mycophagous flagellates	Adl and Gupta (2006); Chakraborty and Pussard (1985); Duczek (1986); Ekelund and Ronn (1994); Ekelund (1998); Hekman et al. (1992); Ogden and Pitta (1990)
Mesofauna
Oribatida	Euzetes globulus Carabodes femoralis Nothrus silvestris Oribatula tibialis	Mitchell and Parkinson (1976); Maraun et al. (2003a); Koukol et al. (2009)
Collembola (=springtails)	Folsomia candida, Folsomia fimetaria, Hypogastrura assimilis, Proisotoma minuta	Dromph (2001); Broza et al. (2001); Dromph and Vestergaard (2002)
Enchytraeidae	Enchytraeus crypticus	Dash et al. (1980); Jaffee et al. (1997); Kristufek et al. (2001); Maraun et al. (2003a); Vänninen (1996); Vänninen (1996)
Macrofauna	Earthworms, Diplopods, slugs and snails	Maraun et al. (2003a)

In the sections below, we review the impact of selected organisms that feed or antagonize fungal BCAs.

Nematodes and oribatid mites

In soil, the most common fungivorous nematodes include species belonging to the genera Aphelenchus, Aphelenchoides, Ditylenchus and Tylenchus (Freckman and Caswell 1985). Nematodes use their stylets to puncture the cell walls of fungal hyphae and withdraw the nutritious cytoplasm. Fungivorous nematodes are usually polyphagous, feeding on a wide range of species of soil fungi including saprophytic, pathogenic, beneficial and mycorrhizal fungi (Freckman and Caswell 1985; Giannakis and Sanders 1989; Ruess, Garcia Zapata, and Dighton 2000; Hasna et al. 2007). Indeed, fungivorous nematodes can reduce the efficacy of beneficial fungi as has been demonstrated for the mycoparasite Trichoderma harzianum (Bae and Knudsen 2001).

Oribatid mites are widespread soil inhabitants that play an important role in recycling leaf litter. They also feed on other organisms such as algae, fungi, collembola and nematodes. Common mycophagous species include: Adoristes ovatus, Eniochthonius minutissimus, Eueremaeus silvestris, Nothrus silvestris, Oppiella subpectinata, Porobelba spinosa and Spatiodamaeus verticillipes (Mitchell and Parkinson 1976; Koukol, Mourek, Janovský, and Cerná 2009). Although there is little evidence of these organisms feeding on EPF, they are known to disperse them (Renker et al. 2005).

Protozoa

Soil protozoa (ciliates, flagellates, amoebae) are major predators of soil bacteria and fungi (Ekelund and Ronn 1994; Ekelund 1998; Adl and Gupta 2006), however, little is known about their impact on fungal BCAs. Since many of the protozoa are polyphagous, it is highly likely that they feed on the conidia and hyphae of EPF. Mycophagous amoebae feed on hyphae and spores (e.g., Thecamoeba granifera, Trichamoeba mycophaga). Few show preferences such as the testate amoeba Phryganella acropodia which, when exposed to Aspergillus niger, Cunninghamella echinulata, Penicillium echinulaturn and Stilbella bulbicola, only ate and digested P. enchinulatum and S. bulbicola (Ogden and Pitta 1990). The vampyrellid amoebae (also called spore perforating forms) attach to the surface of fungal hyphae or spores and, with the aid of hydrolytic cell wall degrading enzymes, enter the cell and digest the hosts cytoplasm. Whereas some species are found in certain habitats, others are more widespread and play an important role in suppressing plant pathogenic fungi (Duczek 1986).

Soil amoebae usually work in concert with other organisms in killing fungal propagules. For example, Fargues, Reisinger, Robert, and Aubart (1983) found that blastospores of B. bassiana incubated in soils were inactivated within 3 weeks. Ultrastructural studies revealed that the blastospores had perforations made by amoebae and bacteria. There was also evidence of physical damage resulting from feeding by mites or collembolans. Formulating the blastospores in a coating of clay substantially retarded the process of biodegradation (Fargues et al. 1983).

Collembola

Collembolans, also known as springtails, occur widely in a range of habitats. Some species are considered plant pests but most are harmless micro-arthropods that play a vital role in recycling, usually feeding on plant residues and soil fungi. Brownbridge (2006) examined the effects of B. bassiana GHA strain (commercialized as BotaniGard™ and Mycotrol™) and M. anisopliae var. acridum (IMI strain; Green Muscle®) on Folsomia candida. No adverse effects on survival or reproduction were detected when Collembola were directly exposed to fungal spores in feeding assays, via contaminated soil. However, Collembola are occasionally infected with EPF, mostly members of the entomophthorales (Brownbridge and Glare 2007). Collembola are reported to feed on EPF and partake in their dispersal (Broza, Pereira, and Stimac 2001; Dromph 2001). Dromph and Vestergaard (2002) showed that three common collembolans, Folsomia fimetaria, Hypogastrura assimilis and Proisotoma minuta, were attracted to conidia of M. anisopliae, B. bassiana and B. brongniartii. These workers found that the order of attractiveness was similar for the three collembolan species with B. brongniartii being the most preferred and B. bassiana as the least. Furthermore, when exposed to a gradient of conidial concentrations, the highest numbers of collembolans were recorded at the highest concentrations of all three fungi.

Enchytraeids and earthworms

Enchytraeids (Oligochaeta, Enchytraeidae) are important and abundant soil inhabitants, especially in peatlands and coniferous forests. The feeding biology of enchytraeids is poorly understood as is their effect on EPF. Enchytraeids are known to feed on fungal propagules in the soil and contribute to suppression of nematophagous fungi (Jaffee, Santos, and Muldoon 1997). Dash, Nanda, and Behera (1980) studied the feeding preferences of three species of enchytraeids and noted that they ingested large amounts of organic material and fungi with fungal material constituting 30% of the gut contents. Analysis of the gut contents showed that Rhizopus nigricans, Syncephalastrum racemosum, and Trichoderma viride were digested, whereas Penicillium steckii and Aspergillus niger were not. This assumption was made because isolations made from postclitellar gut content did not generally give rise to fungal colonies, except for Aspergillus niger and Penicillium steckii. Kristufek, Novakova, and Pizl (2001) reported a 74-fold increase in the numbers of enchytraeids when fed a mixture of Aspergillus flavus and Verticillium tenerum mycelia.

Earthworms play a crucial role in recycling and improvement of soil structure and fertility and inevitably ingest fungal propagules together with earth and organic matter. Conidia are probably digested more due to general feeding as opposed to a preference for the fungal propagules. All laboratory and field studies to date show that EPF are not harmful to earthworms (Brownbridge and Glare 2007; Zimmermann 2007a, 2007b).

Micro-organisms

Micro-organisms can influence the germination, survival, growth and efficacy of fungal BCAs (Clerk 1969; Walstad, Anderson, and Stambaugh 1970; Pereira et al. 1993; Inglis et al. 2001). Whipps (2001) lists some of the microbes that are known to be antagonistic to fungi. Microbes act directly or indirectly on soil fungi. Indirect interactions involve competition for nutrients and niches (e.g., rhizoplane) or parasitization of fungal propagules (Whipps 2001). Direct interactions are due to parasitism and antibiosis. The latter include a wide array of diverse compounds which are secreted or released as volatiles to reduce competition or deter antagonistic organisms which include other microbes (Whipps 2001; Kai, Effmert, Berg, and Piechulla 2007; Vespermann, Kai, and Piechulla 2007). Not surprisingly, some antagonists have been or are being developed for the control of harmful microbes such plant pathogenic fungi (2004). For example, Pseudomonas fluorescens and Streptomyces griseoviridis produce a wide range of antibiotics with antifungal properties and are active ingredients of the fungicidal BCAs Dagger G™ (Ecogen Inc.) and Mycostop (Kemira Oy), respectively. The antibiotics produced by these bacterial BCAs presumably may also affect the efficacy and survival of EPF.

Studies with Beauveria bassiana suggest that fungistasis in natural soils is caused by antibiosis rather than a result of microbial competition. Shields, Lingg, and Heimsch (1981) identified patulin, a water-soluble metabolite of Penicillium urticae, which was inhibitory to conidial germination and growth of B. bassiana. However, many other compounds produced by fungal and bacterial agents are known to be fungistatic. For example, Zou, Mo, Gu, Zhou, and Zhang (2007) found that 328 of 1018 bacterial isolates tested produced antifungal volatiles that could inhibit spore germination and mycelial growth of the nematophagous fungi Paecilomyces lilacinus and Pochonia chlamydosporium. Molecular studies revealed that the bacteria belonged to the Alcaligenaceae, Bacillales, Micrococcaceae, Rhizobiaceae and Xanthomonadaceae. These workers suggested that the volatile compounds acetamide, benzaldehyde, benzothiazole, 1-butanamine, methanamine, phenylacetaldehydeand 1-decene were likely to play important roles in soil fungistasis. Popowska-Nowak et al. (2003) reported that volatile metabolites produced by the actinomycete Streptomyces flavescens totally inhibited growth of all studied strains of B. bassiana, Paecilomyces fumosoroseus and P. farinosus while those produced by the bacteria Bacillus subtilis, Bacillus pumilus and Pseudomonas aurantiaca had only a fungistatic effect. There is increasing evidence that soil microbes produce a wide spectrum of fungistatic volatile organic compounds (Popowska-Nowak et al. 2003; Xu, Mo, Zhang, and Zhang 2004; Kai et al. 2007; Vespermann et al. 2007).

In addition to antibiosis, some microbes will physically attack soil fungi causing hyphal lysis. For example, the mycoparasite Trichoderma harzianum will coil around hyphae of susceptible fungi (mostly plant pathogens) before killing them (Almeida, Cerqueira, Silva, Ulhoa, and Lima 2007). Some actinomycete bacteria, such as Streptomyces griseoviridis, will kill some fungi through hyphal lysis (Tapio and Pohto-Lahdenperä 1991; El-Tarabily and Sivasithamparam 2006).

BCAs of numerous insect pests and fungal pathogens exist, but virtually nothing is known about their interaction if used simultaneously. Krauss, Hidalgo, Arroyo, and Piper (2004) demonstrated differential susceptibility of B. bassiana, M. anisopliae and P. fumosoroseus to the broad host-range mycoparasites Clonostachys spp. and Trichoderma harzianum. In vitro tests revealed that M. anisopliae was highly susceptible to both Clonostachys (formerly Gliocladium) spp. and T. harzianum whereas B. bassiana was attacked by Clonostachys rosea and P. fumosoroseus was resistant to these mycoparasites. Krauss et al. (2004) also showed that co-application of mycoparasites with entomopathogens did not affect their biocontrol efficacy in vivo, and concluded that they were compatible elements of integrated pest management. Other workers have reported that M. anisopliae is compatible with mycoparasitic fungi like Trichoderma and in fact, the two may work efficaciously in the control of some pests (Lopez and Orduz 2003; Butt unpublished observations).

Agricultural (cultural) practices

Tillage

Tillage has a profound affect on the soil biota and is often carried out to reduce the levels of harmful arthropod pests and plant diseases. It can also influence the survival of EPF and explains why the diversity and occurrence of EPF in arable soils is lower than in less disturbed habitats such as permanent pasture, forests, orchards and hedgerows (Gilligan 1990; Chandler, Hay, and Reid 1997; Bidochka et al. 1998; Allsopp, Fischer, Bade, and Dall 2003; Asensio et al. 2003; Bruck 2005; Meyling and Eilenberg 2006a ,b; Sun and Liu 2008).

Generally, more CFUs of EPF are detected per gram of soil from no-tillage systems compared to soils exposed to conventional tillage systems such as ploughing, chiselling and disking (Bing and Lewis 1993; Sosa-Gomez and Moscardi 1994; Hummel et al. 2002). Gaugler, Costa, and Lashomb (1989) on the other hand, observed that application of B. bassiana followed by tillage was very important in achieving enhanced fungal persistence in the soil. The same effect was found by Storey, Gardner, and Tollner (1989). Incorporation of conidia into the soil may protect the inoculum from high temperatures or harmful UV radiation (Hummel et al. 2002). In no-tillage practices, the persistence of EPF in the surface layer of the soil may be enhanced due to protection given by vegetation or availability of susceptible hosts on which to propagate (Hummel et al. 2002; Meyling and Eilenberg 2007).

Crop

The fact that EPF have been recovered from soils from quite diverse habitats/vegetation types (forest, hedgerows, pasture, arable crops) suggests that the vegetation type does not influence their distribution. Sun and Liu (2008) recovered EPF (mostly P. farinosus, B. bassiana and M. anisopliae) from soils collected from forest habitats from different parts of China. Pilz et al. (2008) found a high density of EPF in Hungarian soils with M. anisopliae being detected in every maize field they examined. Sookar et al. (2008) isolated EPF from 38.6% of the soil samples recovered from 19 different locations in three climatic zones in Mauritius with M. anisopliae being the most abundant followed by B. bassiana then P. fumosoroseus. These workers noted that M. anisopliae was isolated more frequently from soils under vegetables as compared to soils under sugarcane or the habitat with natural vegetation (Sookar et al. 2008).

Cropping practices will influence the survival of fungal inoculum. The crop type will influence the pest complex and agronomic practices associated with that crop which, in turn, will influence which strains of EPF become established. Plant topography and growth can influence the efficacy of EPF (Inyang et al. 1998, 1999a; Ugine, Wraight, and Sanderson 2007). Plants can have a direct impact on microbes through the production of antimicrobial compounds. Plant compounds have been identified which interfere with the germination, growth and efficacy of EPF (Lacey 1998; Tallamy et al. 1998; Inyang, Butt, Doughty, Todd, and Archer 1999). Antimicrobial metabolites are known to be produced by plant roots as part of a defense response to pathogenic fungi (Bais, Prithiviraj, Jha, Ausubel, and Vivanco 2005; Bais, Weir, Perry, Gilroy, and Vivanco 2006) but little is known about the effect of these compounds on EPF. For example, Bais, Walker, Schweizer, and Vivanco (2002) found that sweet basil (Ocimum basilicum) root hairs exuded rosmarinic acid in response to plant pathogens like Pythium ultimum. Rosmarinic acid has potent antimicrobial activity against an array of soil-borne micro-organisms (Bais et al. 2002). Many other antimicrobial metabolites have been detected in plant root exudates (Bais et al. 2006).

Metarhizium conidia appear to persist better in the rhizosphere (Hu and St. Leger 2002; Bruck 2005; Wang, Hu, and St. Leger 2005) presumably stimulated by root exudates. However, rhizosphere biota will vary between plant species and could influence the survival and efficacy of EPF (St. Leger 2008).

To date, this phenomenon has only been observed against plant pathogenic fungi. For example, Berg et al. (2002) investigated the effect of plant species on the abundance and diversity of rhizobacteria isolated from potato, oilseed rape, and strawberry and from bulk soil which showed antagonistic activity towards the soilborne pathogen, Verticillium dahliae. These workers note that the proportion of bacterial isolates with antagonistic activity was highest for the strawberry rhizosphere (9.5%), followed by oilseed rape (6.3%), potato (3.7%), and soil (3.3%). Thus, the abundance and composition of Verticillium antagonists was plant species dependent. Interestingly, the predominant antagonists from the rhizosphere of strawberry was identified as Pseudomonas putida B (69%), while those from oilseed rape were members of the Enterobacteriaceae (e.g., Serratia spp., Pantoea agglomerans) (Berg et al. 2002).

Numerous other interactions involving plants influence the persistence, distribution and efficacy of EPF. For example, plants may host endophytic EPF such as B. bassiana. They may release volatile organic compounds that attract potential insect hosts. When ingested by the insect, they may make the pest susceptible to fungal infection. However, these topics are outside the scope of this article and have been reviewed elsewhere (Kepler and Bruck 2006; Ugine et al. 2007; Vega 2008; Vega et al. 2008, 2009).

Fertilizer and pesticide use

Chemical fertilizers, insecticides, herbicides and fungicides are usually applied in conventional farming practices and will influence the survival and efficacy of EPF. Studies of different farming systems by Klingen et al. (2002) showed a significantly higher occurrence of EPF in soils from arable fields of organically managed farms. In conventional farming systems, synthetic pesticides, particularly fungicides, will impact on inoculum levels in various ways. For example, direct contact may kill fungal spores or interfere with their development (e.g., germination or sporulation on hosts).

Chemical pesticides may eliminate the host insects, preventing propagation of EPF or it may weaken the host insect making it more susceptible to infection by EPF or its competitors. Chemicals can make insects more susceptible to EPF even when used at relatively low doses (Quintela and McCoy 1998; Ericsson, Kabaluk, Goettel, and Myers 2007; Shah et al. 2007a). Chemical insecticides may influence insect behaviour such as reduced preening which prevents removal of spores and increased movement, which increases acquisition of inoculum (Shah et al. 2007a; Butt et al. unpublished observations).

Pesticide residues may influence the survival and efficacy of EPF. Studies by Von Kleespies, Bathon, and Zimmermann (1989) and Klingen et al. (2002) suggest that these compounds, especially fungicides applied against plant pathogens, might also negatively affect the populations of EPF by reducing their pest regulation potential. The effects can be direct (Majchrowicz and Poprawski 1993), or indirect i.e., by killing insect hosts on which the EPF could propagate (Von Kleespies et al. 1989; Klingen and Haukeland 2006). Klingen and Haukeland (2006) provided a detailed review of published studies of effects of chemical pesticides on EPF and nematodes. Their main conclusions were that insecticides and herbicides were not deleterious to fungal growth, whereas fungicides were sometimes harmful.

There have been a number of studies to determine the impact of fungicides on entomogenous fungi, however most have been done on in vitro cultures. For example, Shah et al. (2009) investigated the influence of 15 fungicides (azoxystrobin, bupirimate, captan, fenarimol, fenpropimorph, fosetyl-aluminium, iprodione, kresoxim-methyl, myclobutanil, pyrimethanil, quinoxyfen, sulphur, dinocap, tolylfluanid, fenhexamid) on the germination, growth and virulence of M. anisopliae, B. bassiana, P. fumosoroseus, and V. lecanii and found that their influence on conidial germination was dependant on the type and dose of fungicide tested. Most fungicides retarded conidial germination when used at the recommended rate or above. At lower concentrations, their toxicity declined. All fungicides inhibited mycelial growth of B. bassiana, whereas V. lecanii growth was independent of all evaluated fungicides. Only certain fungicides (fenpropimorph, iprodione, tolylfluanid) influenced mycelial growth of M. anisopliae and P. fumosoroseus. None of the fungicides tested influenced the virulence of B. bassiana and V. lecanii. Only one fungicide (tolylfluanid) significantly reduced M. anisopliae virulence and two fungicides, azoxystrobin and kresoxim-methyl reduced virulence of P. fumosoroseus. These studies clearly showed that certain fungicides have the potential to inhibit in vitro germination of EPF but they do not appear to reduce their efficacy (Shah et al. 2009).

Bruck (2009) focused his studies on the impact of fungicides on M. anisopliae and corroborated some of the findings of Shah et al. (2009). He found that a number of fungicides (thiophanate-methyl, dimethomorph, captan, triflumizole, triflozystrobin, pyraclostrobin, azoxystrobin) inhibited the germination of M. anisopliae conidia in vitro. The fungicides fosetyl-AL, thiophanate-methyl, dimethomorph, captan, quintozene, triflumizole, fludioxanil, triflozystrobin, pyraclostrobin, fludiox-mefanox, iprodione, azoxystrobin, and phosphorus acid/K-salts inhibited mycelial growth in vitro. Only three fungicides (etridiazole, propamocard and mafanoxam) had no significant impact in vitro on spore germination or mycelial growth. Bruck (2009) found that, although a number of fungicides had a detrimental impact in vitro, most had no impact on M. anisopliae populations in bulk soil. The only exceptions were captan and triflumizolet, which had a detrimental impact on M. anisopliae populations in the rhizosphere.

Rachappa, Lingappa, and Patil (2007) conducted in vitro studies to determine the compatibility of 10 fungicides, 27 insecticides and 8 herbicides on the germination, growth and sporulation of M. anisopliae. The fungicides carbendazim, propiconazole, chlorothalonil and hexaconazole were highly toxic (i.e., totally inhibiting growth) while the other fungicides tested inhibited growth between 33 and 83%. Captan and wettable sulphur were the least toxic. Of the insecticides assayed, the organophosphate chlorpyrifos was the most inhibitory followed closely by the chlorinated hydrocarbons, endosulfan and dicofol. Spinosad and imidacloprid were the least inhibitory. Butt and his team found that the efficacy of M. anisopliae for vine weevil and thrips control was enhanced when used with low rates of insecticides such as fipronil, imidacloprid, thiacloprid, and chlorpyrifos (Purwar and Sachan 2006; Shah et al. 2007a and references therein).

Due to the complex interactions and composition of agroecosystems, applications of specific fungicides are not necessarily detrimental to the EPF in the soil. It is clear that growers or other end users can select from a wide range of pesticides for use in integrated pest management programmes.

Climatic factors

Temperature, moisture and UV-radiation can influence survival of EPF (Inglis et al. 2001; Meikle, Jaronski, Mercadier, and Quimby 2003). Relatively low humidities (<90%) impede germination and sporulation (Ibrahim et al. 2002), and as stated earlier, soil mositure can influence the survival of inoculum in a range of soil types. Elevated temperatures (>30°C) can stunt fungal growth and death may ensue if temperatures are sustained above 37°C for any significant length of time. However, EPF are found in countries with much sunshine and where day temperatures can exceed 30°C, suggesting that these organisms are fairly robust. Indeed, the upper thermal limits of EPF are around 37°C for conidial germination (Walstad et al. 1970) and 37–40°C for hyphal growth (Fargues, Ouedraogo, Goettel, and Lomer 1997; Inglis et al. 2001). Presumably, the inoculum is protected in some way from harmful radiation or antagonists as discussed earlier and reviewed by Inglis et al. (2001) and Zimmermann (2007a, 2007b, 2008). Recent studies corroborate earlier work which shows considerable variation in thermotolerance between strains, even from the same geographic region (Rangel, Braga, Flint, Anderson, and Roberts 2004; Rangel, Braga, Anderson, and Roberts 2005; Rangel, Anderson, and Roberts 2008; Li and Feng 2009).

Solar radiation has probably the most profound impact on the persistence of fungal inocula. UV-B (280–320 nm) and UV-A (320–400 nm) are the most detrimental components of natural sunlight which cause inactivation of M. anisopliae conidia within a matter of few hours (Braga 2001a, 2001b, 2001c, 2001d). Exposure to even sub-lethal doses of UV radiation can result in genetic and/or physiological changes that can impair germination and growth rates; and thereby reduce the efficiency of EPF (Zimmermann 1982; Braga 2001a, 2001b, 2001c, 2001d, 2002; Rangel et al. 2004, 2008). Under artificial sunlight, the half-life of M. anisopliae conidia was 1 h 40 min after 24 h incubation and 2 h 45 min at 48 h incubation, i.e., the germination process after irradiation is impaired (Zimmermann 1982). The survival of conidia of 23 isolates of M. anisopliae and 14 of M. flavoviride, irradiated with artificial sunlight decreased with increasing exposure, i.e., exposure for 2 h or more was detrimental to all isolates tested (Fargues et al. 1996). However, species and strains of EPF differ in their UV tolerance (Fargues et al. 1996; Braga 2001d; Inglis et al. 2001). Fargues et al. (1996) found that conidia of M. flavoviride were most resistant followed by conidia of B. bassiana, M. anisopliae and P. fumosoroseus. The sensitivity of fungal inoculum to UV light can be influenced by the growth substrate on which conidia were produced (Rangel et al. 2004). Conidia of M. anisopliae are more sensitive to UV when produced on insects followed by PDAY then rice (Rangel et al. 2004). Generally, exposure of fungi to sublethal stresses (exposure to heat, UV-B, nutritive stress) during mycelial growth can increase tolerance of conidia to UV-B radiation.

Conclusions

In the registration dossier of a BCA, the applicant needs to address several data requirements. One of these data requirements is the persistence of the inoculum in the soil. In this review we collected as many data as possible from the open literature on natural background levels and persistence of applied inoculum of three EPF and the factors influencing persistence were reviewed. Many factors were addressed, including intrinsic, edaphic, biotic, cultural an climatic factors, that attribute to the decrease of fungal densities. The following conclusions can be drawn from this review:

1. The Uniform Principles give the criteria for the persistence but it is clear that greater clarity is needed on what constitutes ‘acceptable background levels’. Regulators need to take into account the heterogeneous occurrence of EPF and the influence of land usage/agronomic practices on their occurrence/persistence.

2. The Directive indicates that elevated BCA levels (plateau concentrations) are acceptable if risk assessment does not indicate a problem. Instead of a rigid time frame during which a concentration need to return to natural levels, guidance is needed on how to decide what poses an unacceptable risk to non-targets.

3. A methodology was developed to determine upper natural background levels.

4. Based on this methodology, the upper natural background levels of M. anisopliae, B. bassiana and B. brongniartii were shown to be similar (all approximately 1000 CFU/g soil) even though the data were generated from soils sampled from different parts of the world.

5. Many factors constrain the persistence of natural and introduced EPF inocula leading to a decline in the levels of EPF over time. The data presented should reassure regulators that this is indeed the case for EPF.

6. This review may help to fill in the data requirements for EPF concerning persistence.

More specifically, this study addresses some frequently asked questions (FAQ). These questions, elaborated upon below, are serious points of consideration and are often asked by various stakeholders, including growers, scientists, companies, regulators and others who partake in the evaluation and risk assessment process. For growers and manufacturers of EPF, it is important to know how quickly will the inoculum decline as it will influence the dose and frequency of application. Regulators want to be reassured that the inoculum will decline to background levels. Researchers want to understand the dynamics of EPF so they can have a better understanding of factors influencing fungal survival and initiation of epizootics in insect populations.

FAQ

Q1. Propagation: can we expect uncontrolled growth?

No, almost never. EPF propagate on susceptible hosts resulting in a temporary rise in inoculum levels but these levels will decline due to the biotic and abiotic constraints outlined above. The rate at which inoculum levels decline depends on a range of factors including frequency of application, pest densities, ecological fitness of inoculum, and agricultural practices/land usage.

Q2. What are the upper natural background levels and which level should be used?

Background levels vary from habitat to habitat and sampling method. Conventional sampling methods (e.g., ‘insect baiting’) probably give an underestimate of natural levels. The level of inoculum will alter depending on land usage, e.g., background levels will decline when habitats are converted from natural or semi-natural habitats (e.g., forest, hedgerow) to cultivated arable. Conversely, background levels will rise when arable land is converted to habitats recognised as promoting biodiversity (e.g., forests). For registration purposes, which background level should be used by the risk assessor? In this review, the calculated upper natural background levels were 1040 CFU/g soil for M. anisopliae, 830 CFU/g soil for B. bassiana and 740 CFU/g soil for B. brongniartii. These numbers are to be treated as estimations and a general numbers of 1000 CFU/g soil can be employed for all three species.

Q3. What are accumulated plateau concentrations and are they realistic? (see Uniform Principles)

For risk assessment purposes, the highest realistic concentration following repetitive (multi-year) applications is the accumulation plateau. Simple first-order kinetic demand that the accumulation will reach such a ‘plateau’; the system would then be in a steady state. Dissipation in a certain interval (e.g., 1 year) is then replenished by the treatment. It is a matter of conventions how this plateau concentration is calculated: it could be the highest concentrations after an application or a time weighted average over several intervals. The kinetics of accumulation may be different for EPF compared to those for chemicals since EPF could (temporarily) proliferate on hosts. The Uniform Principles have formulated the principle that the risk of this plateau should be assessed, unless the plateau is not above the natural background level.

Accumulated plateau concentrations have not been observed for the three EPF of this review. Strasser and co-workers (Strasser 1999; Strasser and Enkerli 2001) never reached accumulated plateau concentrations even after five repeated applications of B. brongniartii. Experiments of Nielsen (2004) with four repeated applications of M. anisopliae did not show accumulated plateau concentrations either. Perhaps the plateau concentration could have been attained after several more applications. Repeated applications of EPF would result in a rapid decline in host numbers (mortality is dose dependent). This would reduce propagation of the fungus (due to lack of hosts) and a further decline due to the biotic and abiotic factors outlined in this review. In conclusion, accumulated plateau concentrations do not seem realistic for EPF.

Q4. What should be done when concentrations do not decline to the upper natural background level?

Uniform Principles do not give details of the time for inoculum levels to decline to natural background levels. This review shows that the applied inoculum levels off to upper natural background levels within 0.5–1.5 years for B. bassiana, at least 4 years for B. brongniartii and >10 years for M. anisopliae. The Uniform Principles demand that ‘a robust risk assessment should be made which indicates that the risk from accumulated plateau concentrations is acceptable’. Although accumulated plateau concentrations are not realistic for EPF, the extended period of time that the inoculum density is above the upper natural background level cannot be ignored. It should therefore be discussed what time frame of persistence can be accepted. In principle, the time frame can be very long when no non-target effects are expected. See Q9 for non-target effects.

Q5. What factors influence persistence?

Key factors influencing persistence of inoculum in the soil were reviewed above. In short:

• EPF levels drop as soon as applications cease.

• Antagonists build up over time as a functional response to the introduced EPF (and hosts) and help the decline of EPF.

• A dramatic drop would be expected if the land were ploughed. Ploughing potentially exposes conidia at the soil surface to harmful UV radiation, it disrupts food webs (i.e., potential hosts on which it could propagate) and dilutes out the inoculum and potential infections because host mortality is dose-dependent.

Q6. Can soils / crops be categorised into different persistence types?

All studies to date show a decline of inoculum levels irrespective of strain, crop, soil or climate, region.

Q7. Can persistence be expressed in a DT₅₀ (time for 50% degradation of inoculum)?

For the degradation of chemicals in soil, a first order non-linear decrease is generally assumed and the decrease is determined as a half life time (DT₅₀). This assumption could also apply to EPF at times that suitable hosts are not available. In presence of hosts, the DT₅₀ could be overestimated as amplification of inoculum takes place in the hosts and it thus takes longer to achieve 50% degradation. Since all studies show that EPF inoculum levels decline with time, this suggests that factors contributing to the degradation of inoculum are acting faster than amplification of inoculum on target and alternate hosts.

In absence of hosts, a first order non-linear regression can predict the DT₅₀. At least four data points should be available to obtain a reliable fit. In addition, the correlation coefficient (r ²) should be higher than 0.8. The data points should not include the first period of amplification of inoculum. Such periods are sometimes observed immediately after application.

Q8. Can the results found for M. anisopliae, B. bassiana, B. brongniartii and V. lecanii be extended to other species?

The pattern of decline of the above species is the same irrespective of strain, location, soil type or type of experiment suggesting that similar patterns of decline can be predicted for other strains and species of EPF like Isaria and Lecanicilium. In other words, the spore death rate and the death rate of germinating hyphe and mycelium is greater than the propagation rate.

Q9. Are chronic non-target effects to be expected if background levels do not reach the original baseline?

From a commercial perspective, persistence is a desirable trait that improves performance whereas EU directives (91/414/EEC and 2001/36/EC) want microbial BCA numbers to return to background levels. Growers may wish to retain high levels of inoculum for several years as part of their prophylactic pest management programme (e.g., to protect long-term pasture or fruit crops like blackcurrants, blueberries). The question of non-target effects is of great importance because the relation between the data requirement of persistence and non-target effects is evident.

Persistence itself, even at elevated levels, is not a problem when non-target effects are absent. In the case of M. anisopliae, applications result in maximally 4 log units above the background of 1000 CFU/kg soil. The risk assessment should verify if this is detrimental to non-target species. But, before asking for non-target studies, in-depth reviews focussing on non-target effects on soil organisms, insects, bees, birds and mammals should be performed which must give a clear overview of data available in the open literature. Such reviews, similar to this one, could prevent costly and unnecessary studies.

The effects of EPF on non-target invertebrates and microbes have been reviewed by Goettel and co-workers (Goettel, Poprawski, Vandenberg, and Roberts 1990; Goettel 2001), Draganova, Donkova, and Georgieva (2008) and Vestergaard, Cherry, Keller, and Goettel (2003), respectively. Reviews on non-target effects in the soil are made by Boland and Brimner (2004) and Winding, Binnerup, and Pritchard (2004). Preliminary results from a meta-analysis on the non-target effects of microbial BCAs (not only EPF) on (microbial) soil organisms show that effects are transient (Scheepmaker unpublished).

The fact that host mortality is dose-dependent (Butt and Goettel 2000; Inglis et al. 2001; Butt 2002) corroborates the current opinion that effects are transient. For instance, the effect of M. anisopliae on the larval sugar beet root maggot Tetanops myopaeformis (Jaronski 2007) is only present at high inoculum levels. Once inoculum levels were lower than the target amount of inoculum in the soil, the efficacy sharply decreased. Although T. myopaeformis, in this example, is a target insect species, it can be expected that effects on non-target species decrease similarly.

The decision whether persistence for a longer period of time is acceptable, should ultimately be made by the risk managers.

Concluding remarks

It is clear from this review and previous studies (e.g., Jaronski, Goettel, and Lomer 2003) that there is a need for an urgent review of current data requirements for the registration of EPF being developed for pest control. We reiterate an earlier point that similar reviews could be helpful to aid regulators in making informed decisions as regards the safety of EPF or any other microbial plant protection product. Urgent matters to be coped with immediately are metabolites and toxins produced by microbial BCAs and non-target effects on insects. All stakeholders should encourage these reviews which not only analyse the scientific data but address pertinent questions raised by stakeholders.

Acknowledgements

The authors thank Willem Ravensberg (Koppert), Adi Cornelese (Dutch Registration Authority, Ctgb), Renske van Eekelen (Ctgb) and Mark Montforts (RIVM) for critical review of the manuscript. Jacqueline Scheepmaker thanks RIVM for funding this project from its strategic program.

Notes

1. Repealed in June 2010 by Regulation (EC) No. 1107/2009.