Importance of wild host plants for stem borer parasitoid diversity and control of parasitism in cereal cropping ecosystems in Kenya

Abstract

Field studies were carried out over 2 years (2005–2007) at four locations in Kenya to assess the effects of three plant families (Poaceae, Cyperaceae and Typhaceae) on parasitoid diversity and stem borer parasitism in natural habitats. Field collections were made using destructive sampling. Per locality, stem borer abundance was highest in Poaceae (>69%). Stem borer density per plant (<0.081) did not vary amongst plant families within each locality. Stem borer parasitism was highest in Poaceae in most localities (0.49–2.84%). Although stem borer diversity (Shannon index (H′)) (H′ = 0.09–2.47) did not vary among plant families per locality, stem borer diversity and dominance were positively correlated with parasitism in Poaceae and Cyperaceae. Present results show that Poaceae, Cyperaceae and Typhaceae are important plant families for sustaining parasitoid diversity or function in natural habitats and adjacent cultivated habitats within the cereal crop ecosystem. Wild host plants that harbour parasitoid species or stimulate stem borer activity could also regulate stem borer populations within the cropping ecosystem. Indiscriminate destruction of wild host plants in the Poaceae, Cyperaceae and Typhaceae surrounding cultivated cereal crops should thus be avoided.

Keywords:

• parasitoid communities
• stem borer parasitism
• ecosystem
• natural habitat
• cereal crops
• Poaceae
• Cyperaceae
• Typhaceae

Introduction

Lepidopteran stem borers (in two superfamilies, Noctuoidea and Pyraloidea) generally attack host plants belonging to the families Poaceae, Cyperaceae, Typhaceae and Juncaceae. Only 21 stem borer species attack cereal crops (mainly maize (Zea mays L.), sorghum (Sorghum bicolor L.) and millet (Pennisetum glaucum (L.) R. Br.)) in various parts of Africa (Maes 1997, 1998), while more than 155 stem borer species infest wild host plants (Polaszek and Khan 1998; Le Ru et al. 2006a, 2006b). In spite of lower stem borer species diversity in cultivated habitats, economically important pest species dominate in cultivated cereals, and are the major biological constraint hindering increased cereal production in sub-Saharan Africa. Yield losses due to stem borers vary with region, but generally range from 10% to 80% depending on infestation by the pest species and crop growth stage attacked (Seshu Reddy 1998; De Groote et al. 2003).

In sub-Saharan Africa, cereal crops are usually grown in small fields surrounded by land occupied by wild host plants of lepidopteran stem borers. Some oligophagous/polyphagous (i.e. Chilo partellus Swinhoe, Busseola fusca Fuller and Sesamia calamistis Hampson) cereal stem borers infest one or more wild grass/sedge species (Polaszek and Khan 1998; Mailafiya et al. 2009). Moreover, some oligophagous wild stem borers (i.e. Chilo orichalcociliellus Strand, Busseola phaia Bowden and Sciomesa piscator Fletcher) infest at least one cultivated cereal crop (Polaszek and Khan 1998; Mailafiya et al. 2009). During the off season, when cultivated crops are absent from fields, in addition to the diapausing populations inside crop residues, some cereal stem borers feed actively on wild host plants, from where they infest young crops after planting (Schulthess et al. 1997).

Parasitoids are important natural enemies that generally contribute to curtailing populations of stem borers (Bonhof et al. 1997; Zhou et al. 2003). Without their activities in cultivated cereals, annual crop yield losses would be much higher. In Africa and surrounding islands, approximately 100 genera of indigenous and exotic parasitoids have been recorded that attack stem borer eggs, larvae and pupae (Polaszek 1992, 1998; Bonhof et al. 1997; Kfir 1998). Amongst which, Cotesia spp., Bracon spp., Dentichasmias spp., Chelonus spp., Goniozus spp, Pediobius spp. and Syzeuctus spp. (Hymenoptera) and Siphona spp. (Diptera) are commonly reported (Bonhof et al. 1997; Zhou et al. 2003; Mailafiya et al. 2009).

Insect pests of cultivated crops are increasingly being tackled through managing natural habitats surrounding farmlands to increase agroecosystem biodiversity particularly those of natural enemies including parasitoids and predators (Tscharntke and Brandl 2004; Lindenmayer et al. 2007). Successful habitat management practices, however, require detailed information on the ecological roles of host plants on stem borer parasitoid diversity and their function (parasitisation).

Previous studies in Kenya documented seven stem borer species from only Poaceae plants in cultivated habitats, but 58 stem borer species from Poaceae (48), Cyperaceae (9) and Typhaceae (1) plants in natural habitats (Polaszek and Khan 1998; Le Ru et al. 2006a, 2006b). Parasitoid diversity was further reported to be higher in natural (27) than in cultivated (18) habitats (Mailafiya et al. 2009). Moreover, the majority of parasitoids in cultivated habitats are oligophagous species that attack a narrow range of stem borers across both cultivated and natural habitats. Yet, information on the effects of different host plants on parasitoid diversity and their effectiveness is almost non-existent. Therefore, the goal of the study was to determine the effects of host plant family on parasitoid diversity and stem borer parasitism in four different localities or agroecological zones in Kenya. Knowledge obtained should provide insights into the ecological role of host plants in sustaining parasitoid diversity and their activities against stem borer pests in the cereal cropping ecosystem.

Materials and methods

Survey site descriptions

Field sampling of stem borers and associated parasitoids was carried out from December 2005 to December 2007 at four locations situated in different ecological regions of Kenya, representing different altitudes and climatic conditions (Corbett 1998), as well as vegetation mosaics (White 1983). Descriptions of the study sites are presented in Table 1.

Table 1. Summary description of four different study localities in Kenya

					Mean annual
Locality (Region)	Latitude	Longitude	Altitude (m a.s.l.)	Rainfall distribution	Precipitation (mm)	Temperature (min–max °C)	Vegetation mosaic
Kakamega (Western)	0°10′–0°13′ N	34°54′–34°56′ E	1555–1702	Bimodal	1570	12.7–27.1	Guineo–Congolian rain forest type
Mtito Andei (Eastern)	2°39′–2°40′ S	38°10′–38°21′ E	663–851	Single	665	16.8–29.1	Somalia and Masai Acacia bushland/thicket mosaic
Muhaka (Coastal)	4°18′–4°19′ S	39°31′–39°33′ E	21–54	Bimodal	1210	22.0–30.4	East African coastal mosaic
Suam (Rift Valley)	1°11′–1°12′ N	34°46′–34°48′ E	1665–2035	Single	1190	12.5–26.8	Rain forest and secondary grassland mosaic

Data collection

The number of patches of natural habitats sampled was based on the proportion of natural habitats in each locality (Guihéneuf 2004; Goux 2005), and the availability of patches of natural habitat surrounding cereal fields. Across seasons, 21 patches of natural habitat were sampled nine times in Kakamega, 16 patches were sampled eight times in both Mtito Andei and Muhaka and 10 patches were sampled seven times in Suam. During each sampling occasion, host plants in randomly selected patches of natural habitats bordering cereal fields were sampled, where possible up to 100 m from the field edge. Within each patch of natural habitat, all known/potential host plants belonging to the Poaceae, Cyperaceae and Typhaceae (Le Ru et al. 2006a, 2006b) were inspected for infestation symptoms, such as scarified leaves (window panes and pin holes), dry leaves and shoots (dead hearts), entrance or exit holes and frass. All infested host plants encountered were uprooted and dissected to recover stem borer larvae or pupae in the field (Le Ru et al. 2006a, 2006b). All larvae and pupae obtained were transported to the laboratory on artificial diet and moist filter paper (in Petri dishes), respectively, for rearing/observation till the emergence of adult insects.

Stem borer larvae recovered were reared on an artificial diet suggested by Onyango and Ochieng-Odero (1994) in glass vials (2.5 cm diameter × 7.5 cm depth) plugged with cotton wool and kept under ambient conditions in the laboratory (26 ± 1°C; 65 ± 5% RH) until pupation. Parasitoid puparia or cocoons recovered from stem borer larvae or pupae were kept separately in plastic vials (2.5 cm diameter × 7.5 cm depth) until emergence. Adult stem borer or parasitoid specimens recovered were preserved in 70% ethanol. Stem borers and parasitoids were identified to species level where possible, or otherwise to genera or family level.

Statistical analysis

Since it is difficult to capture all aspects of diversity in a single statistic (Magurran 2004), the various factors considered in this study were assessed using four different diversity indices: Margalef's richness index (D _mg), Pielou's evenness index (J), Shannon–Wiener index (H′) and Berger–Parker index (d). Stem borer and parasitoid diversity were described using the D _mg, a measure of the number of species in a community; J, which indicates the relative distribution of individual species in a community; H′, a measure of species richness and equitability in a community; and the d, which presents the proportional importance of the most dominant species (Magurran 1988, 2004). Apart from being widely used in both ecology and biogeography, these indices describe the attributes considered in this study.

Due to the unequal sizes among the data collected, least squares following logistic regression (generalised linear model, PROC GENMOD; SAS Institute 2001) (Collett 1991) were used to analyse the richness (D _mg), evenness (J), diversity (H′) and dominance (d) of plant families per locality. Significance was set at P ≤ 0.05.

In this study, stem borer abundance refers to the number of individual stem borers recovered (Zhou et al. 2001), stem borer density refers to the number of stem borer individuals per plant (Zhou et al. 2001) and parasitism refers to the proportion of parasitised stem borers (Zhou et al. 2003). Stem borer abundance was analysed across plant families per locality using least squares following Poisson regression (McCullagh and Nelder 1989). Stem borer density and percentage parasitism were analysed across plant families per locality using least squares following logistic regression (Collett 1991). All analyses were carried out using PROC GENMOD (SAS Institute 2001).

Furthermore, the effects of stem borer and parasitoid richness, evenness, diversity and dominance (indices) on parasitism in various plant families and their total per locality were analysed using the non-parametric deviance statistic (PROC GENMOD) with a logistic link function to cater for binomial error distribution (Collett 1991). Pearson's correlation analyses were performed to determine the relationship(s) amongst all factors mentioned. However, prior to analyses, proportion data were arcsine-transformed (Zar 1999).

Results

Stem borer abundance, density and parasitism

A summary of number of stem borers and associated parasitoids recovered from various host plant families inspected in the fields is provided in Table 2. The trophic interactions of all parasitoids collected can be found in Mailafiya et al. (2009).

Table 2. Summary of field collections according to host plant family in four different localities in Kenya

		Host plants inspected		Stem borers recovered		Parasitoids recovered
Locality	Plant family	Number of species	Number of individuals	Number of species	Number of individuals	Number of species	Number of individuals
Kakamega	Poaceae	13	13,200	12	589	9	25
	Cyperaceae	8	2400	5	259	6	22
	Typhaceae	1	600	1	1
Mtito Andei	Poaceae	11	7350	9	459	3	8
	Cyperaceae	3	1000	4	31	2	3
	Typhaceae	1	700	2	150	2	3
Muhaka	Poaceae	11	20,350	10	1816	9	58
	Cyperaceae	2	2150	6	19	2	2
Suam	Poaceae	12	5420	10	318	7	10
	Cyperaceae	7	2700	3	39
	Typhaceae	1	300

Per locality, stem borer abundance was significantly higher in Poaceae than in other host plant families (Table 3). With the exception of Mtito Andei, stem borer abundance was lowest in Typhaceae in all localities of occurrence. In each locality, stem borer density per plant was not significantly different amongst all plant families. By contrast, in all localities, stem borer parasitism was significantly higher in Cyperaceae, followed by Poaceae and least in Typhaceae.

Table 3. Stem borer abundance, density and parasitism across host plant families in four different localities in Kenya

		Stem borer
Locality	Plant family	Abundance	Density (per plant)	Parasitism (%)
Kakamega	Cyperaceae (a)	2.15 ± 0.06	3.75 ± 0.69	25.79 ± 9.08
	Poaceae (b)	2.81 ± 0.04	3.57 ± 0.56	24.63 ± 2.59
	Typhaceae (c)	1.19 ± 0.02	6.39 ± 1.01	0.00 ± 0.00
	df	2, 167	3, 141	3, 167
	P-value	<0.0001	ns	<0.0001
	a vs b	<0.0001		<0.0001
	a vs c	<0.0001		<0.0001
	b vs c	0.0006		0.0006
Mtito Andei	Cyperaceae (a)	1.13 ± 0.17	3.47 ± 0.79	1.67 ± 0.13
	Poaceae (b)	3.05 ± 0.04	2.78 ± 0.48	0.59 ± 0.08
	Typhaceae (c)	2.67 ± 0.08	1.68 ± 0.87	0.03 ± 0.001
	df	2, 89	2, 85	2, 89
	P-value	<0.0001	ns	<0.0001
	a vs b	0.0111		<0.0001
	a vs c	<0.0001		<0.0001
	b vs c	<0.0001		ns
Muhaka	Cyperaceae	0.05 ± 0.01	4.91 ± 1.59	1.16 ± 0.12
	Poaceae	2.22 ± 0.02	2.47 ± 0.23	1.14 ± 0.04
	df	1, 217	1, 216	1, 217
	P-value	<0.0001	ns	ns
Suam	Cyperaceae (a)	1.21 ± 0.11	3.41 ± 1.08	0.46 ± 0.15
	Poaceae (b)	2.55 ± 0.06	3.52 ± 0.81	0.22 ± 0.16
	Typhaceae (c)	0.79 ± 0.31	3.94 ± 0.15	0.09 ± 0.16
	df	2, 84	2, 80	2, 84
	P-value	0.0118	ns	<0.0001
	a vs b	0.0053		0.0001
	a vs c	ns		<0.0001
	b vs c	ns		<0.0001
Note: df, degrees of freedom; P-value, probability value; ns, not significant; vs, versus.

Stem borer and parasitoid diversity

Diversity indices were analysed across host plant families in all localities except Suam; this was due to very low and none recovery of stem borers from Cyperaceae and Typhaceae host plants, respectively. Stem borer richness (D _mg) was significantly higher in Poaceae than other host plant families in half (two) of the localities studied (Table 4). Except in Mtito Andei, where the reverse situation was observed, stem borer evenness (J), diversity (H′) and dominance (d) were significantly higher and lower in Cyperaceae and Typhaceae, respectively. Parasitoid richness (D _mg), evenness (J), diversity (H′) and dominance (d) were generally not significantly different among host plant families in each locality (Table 5).

Table 4. Stem borer diversity indices across host plant families in four different localities in Kenya

		Diversity indices
Locality	Plant family	Margalef's richness	Pielou's evenness	Shannon–Wiener	Berger–Parker
Kakamega	Cyperaceae	0.15 ± 0.06	1.12 ± 0.58	1.32 ± 0.61	1.79 ± 0.81
	Poaceae	1.01 ± 0.20	0.42 ± 0.12	0.37 ± 0.09	0.94 ± 0.08
	Typhaceae
	df	1, 18	1, 18	1, 18	1, 18
	P-value	0.0022	ns	ns	ns
Mtito Andei	Cyperaceae (a)	0.13 ± 0.04	1.36 ± 0.47	1.14 ± 0.37	1.71 ± 0.44
	Poaceae (b)	1.26 ± 0.09	0.73 ± 0.011	0.23 ± 0.07	0.86 ± 0.14
	Typhaceae (c)	0.42 ± 0.11	1.94 ± 0.55	2.32 ± 0.29	2.63 ± 0.41
	df	2, 21	2, 21	2, 21	2, 21
	P-value	<0.0001	ns	ns	ns
	a vs b	0.0020
	c vs d	0.0009
Muhaka	Cyperaceae	1.65 ± 0.57	2.19 ± 0.88	2.56 ± 0.23	2.89 ± 0.24
	Poaceae	1.36 ± 0.45	0.23 ± 0.03	0.09 ± 0.02	0.51 ± 0.15
	df	1, 18	1, 18	1, 18	1, 18
	P-value	<0.0001	0.0226	0.0020	0.0296
Suam	Cyperaceae
	Poaceae	1.48 ± 0.46	2.19 ± 0.53	2.47 ± 1.05	2.81 ± 0.44
	Typhaceae
Note: df, degrees of freedom; P-value, probability value; ns, not significant; vs, versus.

Table 5. Parasitoid diversity indices across plant host families in four different localities in Kenya

		Diversity indices
Locality	Plant family	Margalef's richness	Pielou's evenness	Shannon–Wiener	Berger–Parker
Kakamega	Cyperaceae	0.65 ± 0.04	2.61 ± 0.23	2.14 ± 0.97	2.84 ± 0.83
	Poaceae	0.05 ± 0.01	0.87 ± 0.16	0.84 ± 0.11	1.36 ± 0.66
	Typhaceae
	df	1, 18	1, 15	1, 15	1, 18
	P-value	ns	ns	ns	ns
Mtito Andei	Cyperaceae	2.25 ± 1.33	1.94 ± 0.17	2.32 ± 0.23	2.26 ± 0.41
	Poaceae	0.90 ± 0.05	2.09 ± 0.37	1.99 ± 0.12	2.52 ± 0.36
	Typhaceae
	df	1, 14	1, 14	1, 14	1, 14
	P-value	ns	ns	ns	ns
Muhaka	Cyperaceae	0.99 ± 0.04	1.98 ± 0.34	2.03 ± 0.83	2.55 ± 0.17
	Poaceae	0.43 ± 0.05	0.92 ± 0.09	0.62 ± 0.12	1.22 ± 0.69
	df	1, 18	1, 18	1, 18	1, 18
	P-value	0.0586	ns	ns	ns
Suam	Cyperaceae
	Poaceae	0.69 ± 0.05	1.43 ± 0.52	1.42 ± 0.71	1.92 ± 0.92
	Typhaceae
Note: df, degrees of freedom; P-value, probability value; ns, not significant.

Stem borer/parasitoid diversity correlation with parasitism

For both Cyperaceae and Poaceae host families, stem borer evenness (J), diversity (H′) and dominance (d) were positively correlated with parasitism in at least three localities (Table 6). For Poaceae and Typhaceae, stem borer richness (D _mg) was generally not significantly correlated with parasitism in most localities. In contrast, for Cyperaceae, stem borer richness (D _mg) was significantly correlated with parasitism in at least two localities. For all three host plant families together (total), stem borer richness (D _mg), evenness (J), diversity (H′) and dominance (d) were positively correlated with parasitism in at least half (two) of the localities studied (Table 6).

Table 6. Stem borer diversity indices affecting parasitism in different host plant families in four different localities in Kenya

		Plant family
		Cyperaceae			Poaceae			Typhaceae			Total
Locality	Diversity indices	Deviance	df	P-value	Deviance	df	P-value	Deviance	df	P-value	Deviance	df	P-value
Kakamega	Constant	206.63	9		38.64	9					253.19	18
	Margalef's richness	201.57	1	0.0245	36.95	1	ns				253.19	1	ns
	Pielou's evenness	189.79	1	<0.0001	25.18	1	0.0002				250.63	1	ns
	Shannon–Wiener	202.27	1	0.0368	37.69	1	ns				252.98	1	ns
	Berger–Parker	198.61	1	0.0046	34.61	1	0.0446				253.19	1	ns
Mtito Andei	Constant	72.08	7		20.86	7		18.21	7		126.46	21
	Margalef's richness	6.71	1	<0.0001	12.11	1	0.0031	18.12	1	ns	102.23	1	<0.0001
	Pielou's evenness	0.03	1	<0.0001	15.13	1	0.0166	15.97	1	ns	85.44	1	<0.0001
	Shannon–Wiener	5.69	1	<0.0001	15.14	1	0.0167	17.11	1	ns	79.77	1	<0.0001
	Berger–Parker	3.12	1	<0.0001	12.53	1	0.0039	17.86	1	ns	74.73	1	<0.0001
Muhaka	Constant	147.68	9		22.99	9					177.69	18
	Margalef's richness	146.08	1	ns	22.29	1	ns				169.19	1	0.0036
	Pielou's evenness	136.89	1	0.001	15.83	1	0.0074				174.12	1	0.0013
	Shannon–Wiener	137.89	1	0.001	17.56	1	0.0197				162.52	1	<0.0001
	Berger–Parker	136.89	1	0.001	17.52	1	0.0193				162.37	1	<0.0001
Suam	Constant				59.43	8
	Margalef's richness				56.95	1	ns
	Pielou's evenness				53.22	1	0.0421
	Shannon–Wiener				55.61	1	0.0211
	Berger–Parker				50.72	1	0.0032
Notes: df, degrees of freedom; P-value, probability value; ns, not significant. All diversity indices were positively correlated with parasitism.

Due to low collection of stem borers and parasitoids from Typhaceae, correlation analysis was not performed for this host plant family. Although, for Poaceae, parasitoid richness (D _mg), evenness (J), diversity (H′) and dominance (d) were positively correlated with parasitism in all localities, these factors in Cyperaceae were positively correlated with parasitism in only half (two) of the localities (Table 7). For both Cyperaceae and Poaceae host families together, parasitoid richness (D _mg), evenness (J), diversity (H′) and dominance (d) were positively correlated with parasitism in two localities.

Table 7. Parasitoid diversity indices affecting stem borer parasitism in different host plant families in four different localities in Kenya

		Plant family
		Cyperaceae			Poaceae			Total
Locality	Diversity indices	Deviance	df	P-value	Deviance	df	P-value	Deviance	df	P-value
Kakamega	Constant	206.63	9		38.64	9		253.19	18
	Margalef's richness	178.24	1	<0.0001	29.08	1	0.0022	219.67	1	<0.0001
	Pielou's evenness	185.93	1	0.0041	33.49	1	0.0232	238.24	1	<0.0001
	Shannon–Wiener	182.12	1	0.0231	30.31	1	0.0039	233.95	1	<0.0001
	Berger–Parker	171.23	1	0.0011	30.83	1	0.0052	253.19	1	<0.0001
Mtito Andei	Constant	72.08	7		20.86	7		100.51	21
	Margalef's richness	59.73	1	0.0004	16.28	1	0.0322	96.88	1	0.0566
	Pielou's evenness	60.57	1	0.0021	14.65	1	0.0083	79.56	1	0.0009
	Shannon–Wiener	54.29	1	0.0032	17.22	1	0.0005	87.65	1	0.0003
	Berger–Parker	69.22	1	0.0024	16.64	1	0.0389	85.17	1	<0.0001
Muhaka	Constant	147.95	9		20.53	9		177.69	18
	Margalef's richness	147.82	1	ns	4.69	1	<0.0001	176.52	1	ns
	Pielou's evenness	139.54	1	ns	8.95	1	0.0001	175.33	1	ns
	Shannon–Wiener	141.33	1	ns	2.87	1	<0.0001	176.61	1	ns
	Berger–Parker	146.88	1	ns	3.71	1	<0.0001	176.77	1	0.0382
Suam	Constant				59.43	8
	Margalef's richness				51.21	1	0.0041
	Pielou's evenness				50.55	1	0.0009
	Shannon–Wiener				51.66	1	0.0039
	Berger–Parker				55.37	1	0.0021
Notes: df, degrees of freedom; P-value, probability value; ns, not significant. All diversity indices were positively correlated with parasitism.

Discussion

The fact that stem borer and parasitoid diversity were not significantly different between host plant families in most localities, suggests that stem borer and parasitoid diversity are generally high in the plant families considered in this study. These results are congruent with earlier reports. Le Ru et al. (2006a, 2006b), in assessing the diversity of lepidopteran stem borers on monocotyledonous plants in eastern Africa and the islands of Madagascar reported 135 species of stem borer collected from 75 species of wild host plant belonging to the Poaceae, Cyperaceae and Typhaceae. This, of course, is far less than the total of 21 stem borer species reported attacking four cultivated cereals of the Poaceae: maize, sorghum, millet and sugarcane (Saccharum officinarum L.) in Africa (Maes 1997). Mailafiya et al. (2009), working in Kenya, recorded higher parasitoid diversity on stem borers from wild host plants of the Poaceae, Cyperaceae and Typhaceae (27 parasitoid species) than from cultivated cereals (18 parasitoid species). Also, in a related study involving other herbivore species, Kruess (2003) reported a higher parasitoid diversity on creeping thistle (Cirsium arvense L. Scop.) in non-crop habitats over that in adjacent cultivated crops. Higher richness or diversity of parasitoids in natural habitats might be explained by the utilisation of various alternative herbivore host species in non-crop habitats (Thies et al. 2005). This is particularly true, because cereal stem borer pests are not common on wild host plants in natural habitats (Matama-Kauma et al. 2008; Mailafiya et al. 2010b, 2011). On the contrary, various parasitoid species attack one or more herbivore hosts in different habitat types. Of the parasitoid species recovered in the forest zone of Cameroon by Ndemah et al. (2001, 2007) 37.6% and 41.9%, respectively, were common to both cultivated (on maize) and natural (Pennisetum purpureum Schumach., Setaria megaphylla (Steud.) T. Duran & Schinz or Sorghum arundinaceum (Desv.) Stapf) habitats. Similarly, of all the parasitoid species recovered from different ecological regions in Kenya, Mailafiya et al. (2009) observed 36.4% of the species to be common across cultivated (on maize and sorghum) and natural (on Cyperus papyrus L., Panicum maximum Jacq., P. purpureum, Rottboellia cochinchinensis (Lour.) Clayton, Setaria incrassata (Hochst.) Hack., S. megaphylla (Steud.) T. Duran & Schinz, S. arundinaceum or Typha domingensis Pers.) habitats.

The positive correlation of stem borer diversity and dominance with parasitism in Poaceae and Cyperaceae indicates the importance of stem borer diversity or dominance in these plant families to support parasitoid attacks in natural habitats. Additionally, the positive correlation of parasitoid diversity and dominance with parasitism in Poaceae (in each locality) and Cyperaceae (in most localities of occurrence) showed the importance of these two plant families in maintaining parasitoid diversity and function in natural habitats. Lack of positive correlations as above for Typhaceae, despite high stem borer recovery (abundance) (i.e. from Mtito Andei), was attributed to a combination of low parasitoid diversity (richness, ≤ 2) and parasitism of stem borers in this plant family. Typha plants, particularly T. domingensis, usually occur in swampy and waterlogged areas. Perhaps, many stem borer hosts (either parasitised or unparasitised) die due to drowning during flash floods, after very heavy rains. Heavy rainfall directly inflicts considerable mortality on stem borer populations (Schulthess et al. 2001; Ndemah et al. 2003; Mailafiya et al. 2009). This very likely explains why, during field sampling, numerous dead stem borer larvae and pupae were recovered inside T. domingensis in Mtio Andei (mid-altitude (589–760 m a.s.l.) and rainfall (665 mm)). In fact, at much higher altitudes in Kakamega (1555–1702 m a.s.l. and 1570 mm rainfall) and Suam (1665–2035 m a.s.l. and 1190 mm rainfall), where T. domingensis also occurs, not a single stem borer larva pupa and associated parasitoids was recovered from this host plant species.

Altogether, our results do not support the suggestion of Nye (1960) and Seshu Reddy (1985), encouraging the removal and destruction of wild host plants of lepidopteran stem borers in the vicinity of cultivated cereals as a measure for checking crop infestation, their suggestion clearly did not consider the impact of natural vegetation or habitat destruction on natural enemies of stem borers. Natural habitats, as a component of the ecosystem, provide refuges requisite to parasitoid survival (Godfray 1994; Thies et al. 2005; Wajnberg et al. 2008), and thereby serve as reservoirs for maintaining stem borer parasitoid diversity (Mailafiya et al. 2009, 2010a). Also, by attacking several herbivore host species in both cultivated and natural habitats, various stem borer parasitoid species can change between habitats so as to parasitise hosts, when host availability alternates either across seasons or habitat types (Mailafiya et al. 2009, 2010a). For instance, in the forest zone of Cameroon in Central Africa, parasitoids of cereal stem borers were found attacking four non-economically important stem borer species in natural habitats (Ndemah et al. 2001, 2007). Similarly, in parts of East Africa, parasitoids of cereal stem borers were found attacking 3 and 6 or 19 non-economically important stem borer species in natural habitats in Uganda (Matama-Kauma et al. 2008) and Kenya (Polaszek and Khan 1998; Mailafiya et al. 2009), respectively. Non-economically important stem borers and their host plants thus need to be conserved to at least contribute in maintaining parasitoid diversity within the cereal crop ecosystem. This is particularly important because stem borer density is generally low across the host plant families studied. Moreover, this does not conflict with reports from earlier studies. First, Mailafiya (2009), based on field observations in different ecological zones of Kenya, documented the main sources of infestation of cultivated cereals to be from one or a combination of the following: (1) actively feeding stem borers in irrigated fields (i.e. in Mtito Andei), (2) diapaused stem borers in root stalks within the soil (i.e. in Suam) and (3) in harvested stalks, stashed (i.e. in Suam and Kakamega) or left standing in the fields (i.e. in Mtito Andei and Muhaka). Second, natural habitats seem to be a negligible or minor source of cereal crop infestation by stem borers in parts of East Africa. Evidence provided in Mailafiya et al. (2010b, 2011) on seasonal variability of major cereal stem borers from four different ecological regions in Kenya revealed that S. calamistis and B. fusca are scarce in wild host plants (in the Rift Valley and Western regions, respectively), especially during the dry season. Chilo partellus was, however, found in low numbers across seasons, feeding actively or diapaused in mainly one host plant species, S. arundinaceum, in the Eastern and Coastal regions of Kenya. Similarly, in Uganda, Matama-Kauma et al. (2008) investigated the incidence of stem borers and their parasitism on four selected wild grass species (P. maximum, P. purpureum, Pennisetum polystachion (L.) Schultes and S. arundinaceum), and reported that wild grasses and sedges play a minor role in the seasonal persistence of C. partellus and B. fusca (on only S. arundinaceum). Le Ru et al. (2006a, 2006b) also found most cereal stem borer pests in some wild host plants. Unfortunately, however, their study did not consider the seasonal incidence of the pest species in their wild host plant(s).

Conclusion

Although stem borer parasitism is low in Typhaceae compared with other host plant families (Poaceae and Cyperaceae) considered in this study, all three plant families clearly play an important role in the preservation of parasitoid diversity in the cropping ecosystem. These results, therefore, do not support the need to regularly destroy natural vegetation bordering cultivated cereals in order to avert crop infestation by lepidopteran stem borers (Nye 1960; Seshu Reddy 1985). Generalised destruction of host plants of lepidopteran stem borers in natural habitats should thus be discouraged. On the contrary, natural wild plant communities in the cereal cropping ecosystem should be carefully regulated through habitat management practices that allow or encourage natural enemies of pests to thrive and persist.

Acknowledgements

I am grateful to G. Delvare of CIRAD (Montpellier, France), D. Barraclough of the University of KwaZulu-Natal (Durban, South Africa), P. Moyal of IRD (France) and S. Muthenge of the East Africa Herbarium (Nairobi, Kenya) for the identification of Hymenopteran parasitoids, Dipteran parasitoids, adult stem borers and natural plant hosts, respectively. This research was funded by the German Academic Exchange Service (DAAD) through the African Regional Postgraduate Programme in Insect Science (ARPPIS) at ICIPE and the Institut de Recherche pour le Développement (IRD) in Kenya.