Crop damage increases with pest species diversity: evidence from potato tuber moths in the tropical Andes


*Correspondence author. E-mail:


1.  Insect pests in agricultural systems are one of the major causes of damage to crop production and storage worldwide. However, the study of the effect of multiple pests on agricultural productivity has remained largely disconnected from the ongoing debate on how species diversity affects the productivity of ecosystems. The aim of our study is to use information from crop studies to inform the debate on species diversity and ecosystem productivity.

2.  We present the results of an experimental study that manipulated the species richness of three tuber feeding moth species (Lepidoptera: Gelechiidae) at constant larval density. We measured the influence of this manipulation on (1) damage to the economically most important crop in the Andean region, the potato Solanum tuberosum and (2) the performance of the moths as a consequence of feeding rates.

3.  Our results showed that the three pest species together cause more damage to the crop than is predicted from the effects of each pest alone. This resulted in significant increases in pupal biomass and fecundity.

4.  Potential mechanisms to explain our results are (1) more complete resource utilization and thus greater crop damage (feeding complementarity) and (2) negative interactions, where intra-specific interactions are greater than inter-specific interactions.

5.Synthesis and applications. Our findings may have important consequences for integrated pest management in poor tropical countries. Biodiversity in many tropical countries is decreasing rapidly, leading to reductions in ecosystem services such as biocontrol and pollination. At the same time an increasing number of species, many of them agricultural pests, are being introduced by humans. Our results show that the potential complementarity effects among pest species may increase damage to field crops to a larger extent than previously expected. Control strategies to limit the introduction of new pest species are therefore urgently needed in these countries where the daily management of biological resources is largely in the hands of poor rural people and local government staff with limited funding.


Insect pests in agricultural systems are one of the major causes of damage to crop production and storage (Thomas 1999). In tropical countries, these pests are believed to cause losses approaching 60–70%, principally in stored products (Thomas 1999 and references therein; Nwilene, Nwanze & Youdeowei 2008). Agriculture has long faced a considerable challenge from managing several pest species on a single crop. This has resulted in the development of multi-pest research and survey programmes and specific integrated pest management strategies (e.g. Johnson 1990; Willocquet et al. 2008). Surprisingly, the study of the effects of multiple pests on agricultural productivity has remained largely disconnected from the ongoing debate on how species diversity affects the functioning of ecosystems. Whereas the former line of research has focused mostly on the effect of non-additive species interactions in crop consumption (Castella, Dollona & Savary 2005; Davidson, Peairs & Khosla 2007), the latter has concentrated on additive mechanisms, independently of overall abundance, by which species diversity influences resource consumption, with emphasis on terrestrial plant communities (Loreau & Hector 2001; Hooper et al. 2005; Callaway 2007). Both additive and non-additive insect species interactions probably have an impact on plant damage (Kaplan & Denno 2007), but, so far, insights from studies of the diversity–function debate have rarely been applied to understanding the functional significance of herbivore insect diversity in agricultural systems (Hammons et al. 2009).

There are several practical and conceptual reasons for being concerned with insect pest species interactions in agricultural systems. First, since the establishment of agriculture, planted and stored crops have always been infested by multiple pest species (Savary et al. 1994). This is still the rule today for many economically important crops over the world such as rice Oryza sativa L. in Asia (Savary et al. 1994), corn Zea mays L. and sorghum Sorghum bicolor L. Moench in Africa (Le Rüet al. 2006), wheat Triticum aestivum L. in Europe (Daamen & Stol 1994), corn Z. mays in North America (Davidson et al. 2007) and potato Solanum tuberosum L. in South America (Dangles et al. 2008). Secondly, the increased transport of species by humans in recent years has substantially increased the diversity of many local pest communities in agricultural and forest systems (Duyck et al. 2006; Dangles et al. 2008; Preisser et al. 2008). Because many of these new pests are invasive, often free of natural enemies (Colautti et al. 2004) and able to reach high densities (e.g. Paine 2008), the likelihood and potential importance of inter-specific interactions have increased (e.g. Duyck et al. 2006). Thirdly, research on biodiversity and ecosystem functions in managed ecosystems, such as agricultural landscapes, has focused on the effects of biodiversity loss on ecosystem functions and services (e.g. crop variety: Vandermeer et al. 2002; Li et al. 2007; natural enemy diversity, Wilby & Thomas 2002; Cardinale et al. 2003; pollinator diversity, Morris 2003; Kremen et al. 2007). Much less information is available on the functional consequences of the increase in the diversity of some specific groups of organisms that negatively impact ecosystem productivity (but see Simberloff 2006; Bulleri, Bruno & Benedetti-Cecchi 2008).

Here, we present the results of an experimental study that manipulated the species richness of three tuber feeding moth species Phthorimaea operculella Zeller, Tecia solanivora Povolny and Symmetrischema tangolias Gyen (Lepidoptera: Gelechiidae), one of the most damaging crop pest complexes in the Northern Andes. We measured the influence of this manipulation on (1) damage to the economically most important crop in the Andean region, the potato S. tuberosum and on (2) performance of these three moth species (in terms of survival, biomass and fecundity) as a consequence of resource utilization. Our study was motivated by two observations. First, within the last 30 years, these three moth species have been brought together in combinations of two or three species into potato fields of the Northern Andes (Venezuela, Colombia, Ecuador, Peru and Bolivia) through successive introductions from different origins in South and Central America (Dangles et al. 2008). Although taxonomically related, these species are significantly different in body size, life history, autecology and pathogen resistance (see Dangles et al. 2008) which may result in functional differences in their resource utilization and crop damage. Secondly, in the study region, both storage shelters and sacks of potatoes are now commonly infested with different mixtures of the three moth species. However, depending on latitude and altitude, there are strong differences in crop damage observed among regions. Despite the wide number of natural and anthropogenic variables influencing crop damage in the field (Castillo 2005), data from 31 Ecuadorian potato fields suggest higher crop damage rates at sites where two or three moth species are present than at sites where there is only one species (see Fig. 1 and its legend for details). These observations led us to hypothesize that moth species richness may play a role in the damage these pests cause to potato crops, which mainly occurs under storage conditions in Ecuador. The results of our study indicate that the three pest species together increase damage to the crop to a greater extent than is predicted from the added effects of each pest considered separately, with significant consequences for their performance. Our study further revealed that both pupal biomass and female fecundity of the three species were higher in the multispecies treatments than in single-species treatments.

Figure 1.

 Box-whisker plot of the intensity of damage in potato fields at sites with different richness of the three potato tuber moth species in Ecuador (Tecia solanivora, Symmetrischema tangolias, Phthorimaea operculella). The endpoints of both whiskers indicate minimum and maximum values. Circles indicate which observations can be considered outliers. Data are from 31 potato field sites (n = 8, 13 and 10 for the one-, two- and three-species groups respectively) surveyed in the Central Ecuadorian Andes (for details on the study region see Dangles et al. 2008). Crop damage was measured at harvest. It is expressed as the biomass of damaged tubers in five randomly selected plants in the potato field divided by the total number of adult potato tuber moths of the three species collected with pheromone traps during the 3 months preceding harvest date (for more details on methods see Dangles et al. 2008).

Materials and methods

Experimental design

We performed an experiment investigating the effect of manipulated species richness, from one to three moth species, on potato damage. Experiments were conducted under controlled conditions (i.e. in a room with 70 ± 10% relative humidity, 17 ± 1·5 °C and LD 12 : 12 period). The mean temperature of the room corresponded to mean temperatures in storage structures at sites where the three species co-occur (see Dangles et al. 2008). Our controlled experimental design consisted of eight treatments replicated 10 times; three single-species treatments with 12 individuals of T. solanivora, S. tangolias and S. personatum, three two-species treatments (6 + 6 individuals) with T. solanivora and S. tangolias, T. solanivora and P. operculella, S. tangolias, and P. operculella; one three-species treatment (4 + 4 + 4 individuals) with T. solanivora and P. operculella and S. tangolias, and one control treatment to provide information on tuber-mass loss in the absence of larvae. The constant total density of 12 individuals in each treatment allowed us to test for the effect of diversity without density effects (see Jonsson & Malmqvist 2000). This density was representative of that found in infested sacks (range: 8–18 larvae after 20 days of tuber storage, Dangles and Carpio, unpublished data). For each treatment, potato tubers [S. tuberosum L., v. Leona blanca (Solanaceae) which is equally accepted by the three species] of similar size (31·6 ± 6·9 g) were weighed to the nearest 0·1 mg (Balance TU-OI, FA-2104, Mhand, Fuzhou, China) and placed individually in a 250-mL plastic container covered with nylon mesh. Depending on the number of larvae in each treatment, tubers were inoculated with the corresponding number of eggs taken from permanent cultures in our laboratory. Laboratory colonies were replaced every 3 months with hundreds of larvae collected in two to three potato sacks from the Salcedo market (Salcedo, Cotopaxi Province, Ecuador). Eggs were all of the same age, i.e. laid within 24 hours preceding the experiment. Egg mortality was checked every day using a stereomicroscope (LEICA, MZ3, Leica Microsystems, Wetzlar, Germany). After pupation, tubers were dissected to assess larval mortality and tubers were weighed to the nearest 0·1 mg. Data on loss of tuber mass were corrected for loss caused by factors other than larva consumption (primarily evapotranspiration) as assessed in the larva-free controls. Correction for the effects of moth mortality on potato consumption was made by assuming that dead individuals had lived for half the experimental time. Further, the mass of consumed potato per unit pupal mass was used in the analyses to account for among species differences in biomass.

In addition to larva mortality, we measured pupae biomass (to the nearest 0·1 mg) and the fecundity of emerging females in each treatment as proxies for moth performance. Fecundity was measured as total number of eggs laid per female. Pairs of adult moths were placed in cylindrical plastic containers (h = 20 cm, Ø = 15 cm) covered with mesh nylon (1 mm). Dilute sugar solution provided food and water through cotton swabs. In all three species, the sex ratio was c. 1 : 1 whatever the population density. Oviposition generally started 1 day after adult emergence and the majority of eggs were laid within the first 4 days following mating. For each replicate of each treatment, the fecundity (number of eggs laid by female) of all emerged adult pairs was recorded daily (for additional life table data on the three species, see Dangles et al. 2008). All adults lived for >6 days.

Data analyses

The experiment was a nested design with each species identity treatment nested within a species richness treatment (Jonsson & Malmqvist 2000). This design allowed us to separate the effects of species richness and species identity using nested analysis of variance (anova), including all species combinations in each species richness treatment. We further compared the individual effects on loss of tuber mass caused by each of the three moth species in a one-way anova on log-transformed data. Average values of tuber mass loss in the single-species experiments were then used to predict the results in the multispecies treatments; the residuals from comparisons between predicted and observed mass losses were subsequently used in a second nested anova using data from the experiments with two and three species only (Jonsson et al. 2002).

Following Loreau & Hector (2001), we measured the net effect of species richness on three performance variables of moths, larva survival (ΔF1), pupa biomass (ΔF2) and female fecundity (ΔF3), by the difference between the observed values in the mixture treatments j and their expected values under the null hypothesis that there is no additive effect, as follows:


where FOj and FEj are the observed and expected performance values of species i in the mixture treatment j respectively. The influence of species richness and composition on the three performance variables was further analysed with a nested anova. All analyses were performed using r 2.8.1 (


Mean loss of tuber mass in the control treatment was 1·91 g (SD = ±0·29), about 6·0% of the initial biomass, suggesting a good conservation of tuber quality during the experiment (see Cargill, Briik & Forbush 1989). Taking animal biomass into account, we found significant differences in the feeding rate of the three species in the one-species treatments (Fig. 2). On average, S. tangolias had the greatest impact on loss of tuber mass, followed by T. solanivora and P. operculella (one-way anova, d.f. = 2, F = 8·2351, = 0·0016). Loss of potato tuber mass was significantly affected by moth species richness and identity (nested anova; Table 1, Fig. 2). When species were mixed, loss of tuber mass increased significantly with the number of moth species following an exponential decay model (see Fig. 2, R2 = 0·982, Fstat = 676·63, Spearman test, < 0·001). Tuber mass loss was significantly greater in three-species treatments compared with two-species treatments (nested anova, Table 2) when controlling for differences in single species efficiency.

Figure 2.

 Pest species richness (potato tuber moth larvae) vs. damage on stored potato tuber biomass. Larval density in each treatment was constant (n = 12). Tuber damage values were normalized by larval biomass (mg tuber/mg larva) to account for differences in species size. Error bars represent ±1 SD with 10 replicates. Data were adjusted with an exponential decay model (increasing form) as follows: y = a + b × exp−x with a = 68.03, b = −68.06, R2 = 0.985, Fstat = 676.63, < 0.001 (Spearman test) revealing a significant positive effect of pest diversity on crop biomass. Dashed lines represent 95% confidence intervals. St = S. tangolias, Po = P. operculella, Ts = T. solanivora.

Table 1.   Nested anova for tuber consumption, including all moth species identity treatments
Sourced.f.Sum of SquaresMean of squaresFP
Species number22320·61160·358·99<0·001
Species combination43807·0951·848·39<0·001
Table 2.   Nested anova of the effects of number of species (two or three) and species combination on residual tuber mass loss (mg per mg of larva)
Sourced.f.Sum of SquaresMean of squaresFP
Species number1187·01187·0113·500·0008
Species combination277·0738·532·780·0753

We found no significant net effect of diversity on larval survival (Fig. 3a, nested anova, Table 3). In mixture treatments, larval survival was similar among species (one-way anova, d.f. = 8, F = 1·424, = 0·199). In contrast, we found a significant positive net effect of species richness on both pupal biomass and female fecundity (Fig. 3b, c). For both variables, the net diversity effect was significantly greater in the three-species treatment than in the two-species treatments (nested anova, Table 3), although the difference was much greater for pupal biomass than for female fecundity. In general, the effect of species richness was more variable for fecundity than for larval survival and pupal biomass, especially in the three-species treatment where the number of potential adult pairs for each treatment was only two.

Figure 3.

 Net effect of species richness on three performance variables of potato tuber moths: (a) larva survival, (b) pupa biomass, (c) female fecundity. Each plot represents one of 10 replicates. For clarity in the data representation, the three two-species treatments are slightly offset from each other in the figure. Open dots: T. solanivora S. tangolias, Black dots: S. tangolias P. operculella; Grey dots: T. solanivora P. operculella.

Table 3.   Nested anova of the net effect of number of species (two or three) and species combination on larval survival, pupal biomass and female fecundity
Sourced.f.Sum of squaresMean of squaresFP
(a) Larva survival
  Species number10·000360·000370·4860·4902
  Species combination20·003190·001592·1090·1361
(b) Pupa biomass
  Species number12·88612·886131·267<0·0001
  Species combination25·98332·991632·410<0·0001
(c) Female fecundity
  Species number1167·74167·745·65490·0228
  Species combination2265·52132·764·47570·01837


From small-scale experiments to field conditions

Our results suggest that pest species richness can increase damage to crops. However, although laboratory experiments are helpful for understanding the functional role of diversity among competing species, the more complicated impact of diversity on natural (agro)ecosystem productivity is often not entirely predictable from such small-scale experiments (Loreau, Naeem & Inchausti 2002). The complexity of natural systems includes confounding factors such as community evenness (Dangles & Malmqvist 2004), dispersal (France & Duffy 2006) or trophic interactions (Duffy et al. 2007) that are likely to complicate predictions derived from experimental studies. Even though we addressed the question of the effect of pest species richness on crop damage in a small-scale manipulation experiment, we consider that our findings are likely to be, to some extent, valid also under field conditions. It is unlikely that additional trophic levels may qualitatively change diversity effects observed in our experiments because the three moth species have virtually no predators/parasites when they feed inside the tuber and very few (mainly poultry and insectivorous birds) when they pupate and emerge (Castillo 2005). Moreover, a heap of potatoes stored in traditional storage structures in the Andes is not strikingly different from our experimental units; indeed, once recently hatched larvae have entered the tuber they will develop there until they pupate (Dangles et al. 2008). However, we acknowledge that avoidance mechanisms of pre-infested tubers by young larvae or by ovipositing females (currently being investigated by our team) may modify the frequency of co-occurrence of couples of species in the same tuber (see Choh, Uefune & Takabayashi 2008). Another limitation in our experimental design was that we kept constant both the proportion of each moth species and the total larval densities in all diversity treatments (substitutive design). Several studies have shown, however, that evenness of interacting organisms can influence resource use (Dangles & Malmqvist 2004) and that the effects of species richness on function can be density dependent (Griffiths et al. 2008). Although in our case the natural range of larval abundances inside the potato is strongly limited by tuber size (therefore limiting abundance gradient effects), further studies on how potato moth species assemble under both field and storage conditions would be necessary to predict diversity effects more accurately.

Impact of species richness on crop damage

Our study documents a case of synergetic interaction within a complex of tuber feeding moth species. We found that when all three moth species were together, the damage to potato tubers was higher than predicted from the summed impact of each species alone. These results confirm previous experimental studies that have shown the importance of animal consumer diversity on resource consumption (Duffy et al. 2001; Jonsson et al. 2002; Matthiessen et al. 2007). Individual species effects in this study are different from the situation where a single species has exceptionally large effects on process rates (i.e. the `sampling effect’; Huston 1997), as all species were included in each species richness treatment. Instead, certain combinations of species had greater effects on tuber consumption, especially in multispecies treatments where P. operculella was included (Figs 2 and 3), revealing a significant effect of species identity in addition to a species richness effect.

Several mechanisms, such as niche partitioning (Finke & Snyder 2008) and facilitation (Cardinale, Palmer & Collins 2002; Bulleri et al. 2008) have been shown to contribute to increases in ecosystem functioning/production with diversity. In our case, direct observation of the interior of the tubers at the end of the experiment revealed that the three species exploited tuber resources in different ways. The smallest species, P. operculella, appeared to feed on the periphery of tubers, whereas the two larger species, T. solanivora and S. tangolias, both burrowed deep tunnels into the centre of the tubers. Such differences in the spatial distribution of feeding larvae suggest two mechanisms that can explain our results. (1) It allows more complete resource utilization by the larvae and thus greater damage consistent with a case of ‘feeding complementarity’ already reported in aquatic shredding invertebrates (Jonsson et al. 2002), invertebrate predators in crops (Snyder et al. 2006) and herbivorous coral fish (Burkepile & Hay 2008). (2) It can produce more intense intra-specific interactions that would hamper process rates more in species-poor situations compared to where encounters and interactions take place primarily between species (Jonsson & Malmqvist 2000). As predicted by Loreau & Hector (2001), niche complementarity of the three species would potentially enhance the rate of tuber loss beyond that of the best performing single species, which was verified in our experiment. Here, again it is important to record that, assuming constant resource availability, mechanisms driving species richness effects might vary depending on both relative and total species abundance in the assemblage (Griffiths et al. 2008).

Our study further revealed that both pupal biomass and female fecundity of the three species were higher in the multispecies treatments than in single-species treatments. Individual performance (e.g. survival and fecundity) has been shown to be influenced by inter-specific interactions in plant (Callaway 2007; Lortie & Turkington 2008) and animal communities (e.g. Kaplan & Denno 2007 for insects). However, much less is known about the effect of species richness on individual performance in multi-species animal communities, and our study is one of the few manipulative experiments that addresses this issue (see also Snyder et al. 2006 for invertebrate predators). In our experiment, the increase in resource consumption and/or lower intra-specific crowding density of larvae in multi-species treatments may have led to the observed increased pupal biomass and female reproductive performance (Danthanarayana, Hamilton & Khoul 1982; Zheng et al. 1993). These results suggest that pest species diversity has a significant effect not only on crop damage level but also on pest population dynamics in locations where two or three species coexist. Further studies on the performance properties of insect pest populations in complex mixtures of co-occurring species will provide insights not only in our understanding of pest population dynamics but also in the contribution of indirect interactions to the organization of phytophagous insect communities (Kaplan & Denno 2007).

Synthesis and applications

Although crops are almost universally subject to attack by multiple herbivores and pathogens, our understanding of how multiple plant enemies affect each other’s dynamics and damage on plants has not been considered in detail by crop scientists (see Fournier et al. 2006). In this context, the results of this study have two major implications for applied ecologists studying crop management and biological control. First, our results indicate that complementarity effects among herbivores species may increase damage to field crops to a larger extent than previously expected. This finding is of particular concern for poor tropical countries where biodiversity is decreasing rapidly (therefore decreasing ecosystems services such as biocontrol and pollination), while at the same time an increasing number of species (many of them agricultural pests) are introduced by humans at accelerating rates (Mooney & Hobbs 2000). Moreover, the generally warm and/or diverse climatic conditions in tropical countries may favour the population dynamics of introduced pests (see Dangles et al. 2008), which itself can be fostered by positive interactions among species. Instead of focusing on one particular pest species, potato growers should therefore aim to limit the diversity of potato moths by using non-specific management strategies. Because the three pest species have different flight phenologies and attack behaviour in the field (Roux 1993; Castillo 2005), farmers should concentrate their efforts at the storage stage. They could for example use control strategies that physically protect tubers against larva attack, whatever the species (see Pollet, Barragan & Iturralde 2004). At a regional level, reinforced control strategies to limit the introduction of new herbivorous pest species are urgently needed in developing countries where the daily management of biological resources is largely in the hands of poor rural people and local government staff with little operational funding.

Secondly, the evidence of complementary interactions among multiple agents on crops has promising applications in biological control terms, as previously reported for the multiple enemy effects on pest suppression (Cardinale et al. 2003; Snyder et al. 2006). In the specific case of herbivores, our results provide interesting insights in the context of weed control. Although a number of studies have shown that a single species of herbivorous insect can result in a significant reduction of weed biomass (Zwolfer & Harris 1971; Oraze & Grigarick 1992; Sheldon & Creed 1995) our findings indicate that the impact of multiple herbivores can be higher. If the aim of weed management programmes is to maximize the overall impact of herbivory on target plant populations, then combinations of herbivores (or in a broader sense, of agents) exploiting a similar host and interacting synergistically will be especially valuable (see Denoth, Frid & Myers 2002; Davis et al. 2006; Fournier et al. 2006). Moreover, while the present work focused on additive effects within one trophic level, further studies should examine the effects of species diversity effect across trophic levels (see for example Hämback, Agren & Ericsson 2000). This lies at the heart of the present diversity–ecosystem function debate and is likely to have important applications for crop management, especially in poor tropical countries that strongly depend on agricultural resources.


This work was part of the research conducted within the project ‘‘Biopesticide development and diffusion for the control of potato moths’’ (C14-026) funded by the McKnight Foundation. Additional funding was provided by the School of Biology of the Catholic Pontifical University of Ecuador. The authors are grateful to J. Padilla, C. Mazoyer and P. Rosero for technical assistance during the experiment. We are grateful to F. Anthelme, J. Casas, S. Dupas, M. Jonsson and B. LeRü for constructive criticisms and to C. Keil for the linguistic revision of the manuscript. The authors are grateful to the editor and two anonymous reviewers for their helpful comments on a previous version of the manuscript.