Expansion of the Cassava Brown Streak Disease Epidemic in Eastern Democratic Republic of Congo

Cassava (Manihot esculenta Crantz) is an important staple food crop in sub-Saharan Africa. In Central Africa in particular, most people depend on cassava as the main source of carbohydrates (Dunstan and Chuma 2017; Katinila et al. 2002). In the Democratic Republic of Congo (DRC), cassava plays a vital socioeconomic role in smallholder farming systems (Masika et al. 2019; Munyahali et al. 2017). Although cassava is adapted to a wide range of agroecological conditions (El-Sharkawy 2006), it is vulnerable to infection by several plant viruses (Thresh et al. 1994). In Africa, cassava mosaic virus disease (CMD) and cassava brown streak disease (CBSD) are together responsible for losses of >$1 billion US annually (Legg et al. 2006; Ndyetabula et al. 2016). Pandemics of both virus diseases have caused devastating losses to cassava production in East and Central Africa from the 1990s to the present day. The causal viruses are spread by unusually abundant populations of the whitefly vector, Bemisia tabaci (Gennadius) (Aleyrodidae) (Legg et al. 2011, 2014). CBSD used to be geographically limited to coastal eastern and southern Africa. New epidemic spread from the mid-2000s to the present day through Uganda, Burundi, Rwanda, western Kenya, northwestern Tanzania, and eastern DRC (Alicai et al. 2007; Ntawuruhunga and Legg 2007) has heightened concern about the risk of further spread.

CBSD is caused by two species of cassava brown streak ipomoviruses (CBSIs; family Potyviridae; genus Ipomovirus): Cassava brown streak virus (CBSV) and Ugandan cassava brown streak virus (UCBSV) (Mbanzibwa et al. 2009; Winter et al. 2010). CBSIs are perpetuated by planting stem cuttings obtained from infected parent stems and through vector transmission by the whitefly B. tabaci (Maruthi et al. 2005, 2017; Storey 1936; Tata-Hangy and Mahungu 2014). Symptoms of CBSD occur on all parts of the cassava plant, but the most important damage is a necrotic root rot that makes them unsuitable for consumption and marketing (Nichols 1950; Storey 1936). CBSD causes yield loss in susceptible varieties both by reducing the growth as well as through eliciting a dry necrotic rot that develops in storage roots as they mature (Ndyetabula et al. 2016). Symptoms are present on leaves but are much less obvious than those of CMD, as they do not distort the leaf shape and occur mainly toward the bottom of the plant (Nichols 1950). Molecular diagnostic studies have shown that a proportion of plants infected by CBSIs do not show symptoms at the time of sample collection (Adams et al. 2013). This increases the risk that these viruses may be propagated inadvertently through stems used for planting material.

CBSD was reported for the first time from eastern DRC in 2011 (Mulimbi et al. 2012). This first report recorded the presence of UCBSV in Nord-Kivu Province. Casinga et al. (2019) noted the occurrence of both UCBSV and CBSV as well as mixed infections of the two species in Ituri Province, neighboring Nord-Kivu. Reports of CBSD from northern Zambia, in districts bordering DRC (Mulenga et al. 2018), have led to concern about possible CBSD occurrence in southeastern areas of DRC. There were also unconfirmed reports of CBSD in the Ruzizi Plain and shores of Lake Tanganyika in Sud-Kivu Province and there has been concern about further southward and westward spread of CBSD across DRC (United States Agency for International Development 2017). This study aimed to determine the current occurrence of CBSD in eastern DRC through an extensive surveillance of CBSD and CBSIs during two time periods in 2016 and 2018.

Materials and Methods

Survey area.

Assessments of CBSD within farmers’ fields were conducted during two periods in five provinces of eastern DRC. The first assessment was from November 2015 to September 2016 (hereafter referred to as 2016) and the second was from March to August 2018 in Sud-Kivu, Nord-Kivu, Tanganyika, and Haut-Katanga provinces in eastern DRC (Fig. 1). Ituri Province was only surveyed in 2016. The total area of provinces surveyed covers 461,636 km² and is located between 24.75 and 31.66°E longitude and 3.96°N and 13.94°S latitude.

Download as PowerPoint

Assessing fields for CBSD occurrence and whitefly abundance.

CBSD and whiteflies were assessed within 390 cassava farmers’ fields (3 to 6 months old). A total of 201 fields were surveyed in 2016 and 189 in 2018. Geographical coordinates and altitude were recorded for each of the sampled sites with a Garmin Etrex handheld GPS unit. Cassava fields were inspected at 5- to 10-km intervals along motorable roads, although distance often varied depending on the availability of farmer fields. Variety names were obtained and recorded from farmers. In a small proportion of cases, farmers did not know the variety name, in which case these were recorded as “unknown local.” In each sampled field, 30 plants selected randomly along two diagonals (15 plants per diagonal) were assessed for foliar symptoms of CBSD by scoring disease severity using a numbered scale from 1 to 5, where 1 represents no visible symptoms and 5 indicates very severe symptoms (Hillocks et al. 1996). For each diagonal, its distance was estimated, and this value was divided by 15 to give the spacing between sampled plants. All sampled plants were taken from the right-hand side of the transect walk in a systematic way to avoid any bias related to symptom expression. The two diagonals crossed at the center of the field, where care was taken to avoid sampling the same plant twice. Data obtained were eventually used to estimate mean CBSD severity and CBSD incidence values for each of the visited sites. Mean CBSD severity was the average severity score for all plants that were symptomatic (i.e., with a score between 2 and 5). CBSD incidence was the percentage of plants that were symptomatic. B. tabaci whitefly vector abundance was determined in each visited field. To do this, B. tabaci adults were counted on the undersides of the top five leaves of the tallest shoot of each of the 30 assessed plants. Total counts for these plants were averaged to produce a mean whitely abundance value for each field (Sseruwagi et al. 2004). Within each visited field, 10 leaf samples were collected for CBSI reverse transcription (RT)-PCR testing from every third assessed plant along the two diagonals. The central leaf lobe was detached from the fifth fully expanded leaf of the shoot used for counting whiteflies and attached to a newsprint sheet using masking tape and preserved dry in a herbarium press until required for nucleic acid extraction. Collected leaf samples were transferred to and preserved in the laboratory at the International Institute of Tropical Agriculture (IITA)-Kalambo in Bukavu, eastern DRC.

Diagnostics and sequencing of CBSI samples.

Samples were tested from 131 of the 201 surveyed fields in 2016 and from 134 of 189 fields in 2018. These included representative fields from all provinces surveyed. Testing focused more strongly on northern provinces in 2016 and southern provinces in 2018. Individual samples were tested for all fields in which symptoms of CBSD were observed. Where these symptoms were not observed, nucleic acid extracts of the 10 samples were pooled for testing. Total RNA extraction was performed using a standardized cetyltrimethylammonium bromide method (Lodhi et al. 1994; Xu et al. 2010). For 2016, samples were analyzed through a single-step RT-PCR with the primers CBSV10 (5′-ATCAGAATAGTGTGACTGCTGG-3′) and CBSV11 (5′-CCACATTATTATCGTCACCAGG-3′) (Monger et al. 2001), which were used to provide a generic test of the presence of CBSIs. A 25-µl single-step RT-PCR assay was set up containing 1× M-MLV RT buffer (New England Biolabs [NEB]), 200 mM of dNTPs, 200 µM of each primer, 1 unit of Taq DNA polymerase (NEB), 10 units of Moloney murine leukemia virus reverse transcription (NEB), and 60 ng of total RNA. Suitable controls were included for each batch of RT-PCR testing, in which positive controls were previously identified RNA samples and PCR-grade water was used as a negative control. Additionally, a cytochrome oxidase I gene assay was included as an internal control to check for RNA quality (Adams et al. 2013). For species-level confirmation, 10 samples selected from territories in Ituri, Sud-Kivu, and Nord-Kivu provinces were tested using TaqMan real-time quantitative reverse transcriptionPCR (qPCR) (Adams et al. 2013; Shirima et al. 2017) at the IITA Molecular Biology Laboratory in Dar es Salaam, Tanzania. For 2018, samples were tested for CBSIs from 134 sites (both symptomatic and asymptomatic) using TaqMan real-time qPCR at IITA-Kalambo in Bukavu. For the 2016 sample set, 18 CBSI-positive samples were selected for direct PCR product sequencing. These comprised 10 sequences from Aru and Mahagi territories in Ituri Province, five from Rutshuru Territory in Nord-Kivu Province, and three from Uvira Territory in Sud-Kivu Province. Amplicons for this purpose were generated using a two-step RT-PCR method in which 826-bp (CBSV) and 732-bp (UCBSV) RT-PCR products were generated employing the following primer pairs: CBSV-CP-Fer2 (3′-GAAGGGATTGGAYTRGAAGGA-5′)/CBSV-CP-R1-1 (3′-GAACGCGGTATCCACACATA-5′) and UCBSV-CP-F1-1 (3′-AGAGATCTGGAAAGGAAGT-5′)/UCBSV-CP-R1-1 (3′-CTCGCCAYGACTTCTCATT-5′) (Shirima et al. 2017). The amplicon target for CBSV spanned a portion of the coat protein (CP) gene from nucleotide positions 7,390 to 8,216 (GenBank accession GQ329864), whereas that for UCBSV targeted the CP region of UCBSV between nucleotide positions 7,689 to 8,421 (GenBank accession FN434109). Sequences were generated using the standard Sanger dideoxy method at Macrogen Inc. (Rockville, MD) and edited and assembled with CLC Main Workbench version 7 (Qiagen, Venlo, Netherlands).

Data analysis.

QGIS 2.6.1 software (https://qgis.org/en/site/) was used to map the survey locations and maps illustrating CBSD data were produced using ArcGIS 2.16 software (ESRI, Redlands, CA). Variation among sampled variables between provinces and years was investigated using generalized linear mixed model (GLMM) analysis of variance (ANOVA), which was run using SAS software (SAS Institute, Cary, NC). The two-way ANOVA examining both year and province was run excluding the Ituri data since these were only collected in 2016. One-way ANOVA to assess variation among whitefly abundance as well as CBSD incidence was run separately for years 2016 and 2018 using a GLMM approach in SAS. GLMM ANOVA tests for CBSD incidence and whitefly abundance used the Gaussian and Poisson distributions, respectively. Mean severity scores for CBSD-affected fields in different provinces were compared using the nonparametric Kruskal-Wallis one-way ANOVA on ranks procedure in which Dunn’s method was used for pairwise comparisons (SigmaStat; Systat Software Inc., San Jose, CA). A phylogenetic tree was generated for CBSI sequence data using the maximum likelihood procedure with the Tamura-Nei model (Tamura and Nei 1993) in MEGA7.0 with default settings and 1,000 bootstraps (Kumar et al. 2016). Based on the phylogenetic relationship results for sequences, the generated clades were selected to estimate similarities between the group means (Tamura et al. 2007). Eight of these sequences were published in 2019 (Casinga et al. 2019).

Results

CBSD symptoms, incidence, and severity.

Leaf, stem, and root symptoms (Fig. 2) typical of CBSD were observed during field inspections in 11 territories during the two survey periods. Considering only the four provinces sampled in both 2016 and 2018, 13 territories were sampled in both years; of these, five (39%) had fields with CBSD in 2016, which increased to eight (62%) in 2018. There was no significant change in the incidence of CBSD from 2016 to 2018, but there were significant differences in incidence between the provinces based on an analysis of the whole dataset for both years (F = 9.58, df = 3, P < 0.0001) (Table 1). Considering the 2016 data independently, there were significant differences in CBSD incidence between provinces (F = 11.63, df = 4, P < 0.0001). Sud-Kivu and Ituri had higher incidences than both Haut-Katanga and Tanganyika (Table 1). In 2018, there were no significant differences in incidence between provinces (F = 2.36, df = 3, P = 0.073). CBSD was completely absent during both surveys in Tanganyika Province (Figs. 3 and 4). In 2016, the highest incidence was recorded in the most northerly province of Ituri (Fig. 3). CBSD was only present in the northernmost part of Haut-Katanga Province in Pweto Territory (Fig. 3), and overall incidence in this province was very low (1.7%). The most notable change between the two survey periods was the increase in incidence in Haut-Katanga Province, which rose to 15.9% in 2018 (Table 1). Associated with this was a large increase in prevalence (percentage of fields with at least one sampled symptomatic plant) from 6 of 60 (10%) in 2016 to 11 of 36 (31%) in 2018. Significant differences were observed in CBSD foliar severity between provinces in 2016 (Table 1) (H = 22.5, df = 3, P < 0.001) but not in 2018. Symptoms were more severe in Ituri, Nord-Kivu, and Sud-Kivu provinces than they were in Haut-Katanga (Table 1).

Download as PowerPoint

Table 1. Incidence and severity of cassava brown streak disease and whitefly abundance in surveyed fields of cassava in the provinces of eastern Democratic Republic of Congo: 2016 and 2018^y

Province	2016				2018
	Fields (n)	Whitefly abundance	CBSD incidence	CBSD severity	Fields (n)	Whitefly abundance	CBSD incidence	CBSD severity
Haut-Katanga	60	7.41 a	1.67 b	2.03 b	36	0.87 c	15.93 ab	2.97
Ituri	36	8.15 a	30.09 a	3.24 a	NS	NS	NS	NS
Nord-Kivu	12	4.17 b	16.11 ab	3.32 a	29	3.32 a	24.02 a	2.78
Sud-Kivu	52	5.04 b	24.88 a	2.92 a	106	1.75 b	16.76 ab	2.93
Tanganyika	41	2.24 c	0.00 b	ND	18	1.41 bc	0.00 b	ND
Total or average	201	5.68	13.29	2.92	189	1.79	16.12	2.90
Four-province averagez		4.72	10.67	2.76		1.84	14.18	2.89

^yMean values for provinces with different letters were different at the P < 0.05 level. Whitefly abundance is the mean number of adult whiteflies per plant counted on 30 plants in each surveyed field. For each sampled plant, whiteflies were counted on the undersides of the top five leaves of the tallest shoot. ND = no disease and NS = not surveyed.

^zAverage value of the province means for the four provinces that were surveyed in both 2016 and 2018.

View as image HTML

Download as PowerPoint

Download as PowerPoint

Varieties and responses to CBSD.

Eighty-three cassava varieties were recorded during the two survey periods, of which 73 were local and 10 were improved varieties released by the national research system. The 21 most frequently encountered varieties (five or more fields of each) made up 66.4% of all fields (Supplementary Table S1). Only three of the improved varieties were frequently encountered, although improved Sawasawa (MM96/3920) was the second most frequent overall, after Ndelekulwa. All provinces had a similarly high degree of varietal diversity and most of the varieties encountered were confined to one or two territories within a province. The widely varying frequency with which different varieties occurred precluded an analysis of varietal interactions with whitefly abundance, CBSD incidence, and CBSD severity. However, there was no clear difference in the pattern of CBSD infection in improved versus local varieties (Supplementary Table S1). The highest incidence of CBSD was recorded in Ituri Province for variety Nyagota (77.2%). This variety also had the most severe CBSD symptoms (3.62; Supplementary Table S1). Two commonly grown varieties (Mwabayilo and Nyamwalirwa) had no CBSD despite being cultivated in Sud-Kivu Province, in which CBSD was otherwise widespread.

Whitefly abundance.

B. tabaci whitefly adults occurred throughout the sampled regions at moderate to low abundances but there was no consistent regional pattern of abundance (Table 1). Statistical analysis of the data revealed significant differences between provinces (F = 9.86, df = 3, P < 0.0001) as well as between years (F = 122.1, df = 1, P < 0.0001); there were more whiteflies in 2016 than in 2018. In 2016, B. tabaci was most abundant in Ituri and Haut-Katanga provinces and least abundant in Tanganyika. In 2018, abundance was greatest in Nord-Kivu and least in Haut-Katanga. Abundance of B. tabaci whiteflies varied greatly from field to field: 66.2% of fields had mean abundance values of <1, whereas 5.1% had an abundance of >20 adults per top five leaves. The only region that provided an exception to this pattern was the far southeastern portion of Haut-Katanga Province, in which all fields sampled had a low abundance of B. tabaci whiteflies.

Occurrence and phylogenetic relationship of CBSIs in eastern DRC.

PCR testing confirmed the presence of CBSIs for virtually all territories in which CBSD symptoms were observed in both 2016 and 2018 (Table 2). The only exceptions to this were a single field in Djudu Territory in Ituri Province in 2016, where CBSIs were detected in the absence of clear CBSD symptoms, as well as two fields in Kalehe Territory in Sud-Kivu Province, where CBSIs were not detected when CBSD symptoms were observed.

Table 2. Frequency of fields surveyed in 2016 and 2018 in eastern Democratic Republic of Congo in which cassava brown streak disease (CBSD) symptoms were present or absent and for which positive or negative test results were obtained from single-step reverse transcription (RT)-PCR and quantitative RT-PCR (qPCR) diagnostic tests^z

Province	Territory	2016					2018
		CBSD affected	CBSD free	PCR positive	PCR negative	Not tested	CBSD affected	CBSD free	Not tested	qPCR positive	qPCR negative
Ituri	Mambasa	0	6	0	6	0	NS	NS	NS	NS	NS
	Irumu	0	7	0	7	0	NS	NS	NS	NS	NS
	Mahagi	7	8	8	7	0	NS	NS	NS	NS	NS
	Aru	5	2	7	0	0	NS	NS	NS	NS	NS
	Djudu	0	1	1	0	0	NS	NS	NS	NS	NS
Nord-Kivu	Masisi	0	6	0	6	0	5	12	16	1	0
	Bweremana	0	1	0	1	0	NS	NS	NS	NS	NS
	Rutshuru	4	0	4	0	0	7	0	1	6	0
	Nyiragongo	0	1	0	1	0	1	4	2	2	1
Sud-Kivu	Kabare	0	7	0	2	5	1	18	4	1	14
	Walungu	2	3	2	1	2	7	16	12	5	6
	Uvira	9	1	1	0	9	8	1	5	4	0
	Fizi	21	9	7	10	13	12	10	13	9	0
	Kalehe	NS	NS	NS	NS	NS	2	13	0	0	15
	Idjwi	NS	NS	NS	NS	NS	0	18	0	0	18
Tanganyika	Kalemie	0	19	0	19	0	0	18	0	0	18
	Moba	0	22	0	22	0	NS	NS	NS	NS	NS
Haut-Katanga	Pweto	5	27	1	18	13	12	6	2	10	6
	Kasenga	0	17	NT	NT	17	0	5	0	0	5
	Kipushi	0	5	NT	NT	5	0	9	0	0	9
	Sakania	0	2	NT	NT	2	0	3	0	0	3
	Kambove	0	4	NT	NT	4	0	1	0	0	1
Total		53	148	31	100	70	55	134	55	38	96

^zPCR positive indicates fields in which positive test results were obtained with single-step RT-PCR diagnostic tests, PCR negative indicates fields in which only negative test results were obtained with single-step RT-PCR diagnostic tests, qPCR positive indicates fields in which positive test results were obtained with qPCR diagnostic tests, and qPCR negative indicates fields in which only negative test results were obtained with qPCR diagnostic tests. NS = not surveyed and NT = not tested.

View as image HTML

In 2016, both CBSV and UCBSV were detected in Mahagi Territory (Ituri Province), whereas only UCBSV was present in Aru Territory (Ituri Province), Uvira Territory (Sud-Kivu Province), and Rutshuru Territory (Nord-Kivu Province) (Fig. 5). UCBSV was detected in all 10 fields (100%) from which qPCR results were obtained, whereas CBSV was only recorded from two fields (20%) in Mahagi Territory. CBSV occurred only as mixed infections with UCBSV in both fields and not singly. In 2018, single infections of UCBSV were detected in three territories of Nord-Kivu (Masisi, Rutshuru, and Nyiragongo) (Fig. 6). Only single infections of CBSV were detected in Pweto Territory in Haut-Katanga and Kabare Territory in Sud-Kivu, whereas both viruses occurred in Walungu, Uvira, and Fizi territories in Sud-Kivu (Fig. 6). Neither CBSV nor UCBSV was detected in either year in Tanganyika Province. Of the 38 detections in 2018, 28 were CBSV (74%) compared with 13 for UCBSV (34%). There was therefore a large increase overall in the proportion of CBSV-infected fields in 2018 (74%) compared with 2016 (20%), although the number of qPCR tests was much less in 2016. In three fields in Walungu, Uvira, and Fizi, both viruses were detected, but in each case different virus species were present in different plants (i.e., there were no dual infections).

Download as PowerPoint

Download as PowerPoint

Phylogenetic analysis of 12 CBSV and UCBSV CP sequences obtained from Ituri, Nord-Kivu, and Sud-Kivu confirmed that all were closely related to UCBSV or CBSV sequences from locations in East and Southern Africa (Fig. 7). UCBSV sequences from Nord-Kivu (Rutshuru) and Ituri fell within a single large clade that also includes sequences from Uganda, coastal Kenya, coastal Tanzania, and Malawi. All of the UCBSV sequences from the current study shared >99% homology with nearest relatives in GenBank (Fig. 7) with the exception of Uvira553 (MF511057), which was 98.4% identical to a sequence obtained from Mtwapa, coastal Kenya. The two CBSV sequences from this study were identical and fitted clearly within the major CBSV species clade. They were most closely related to a sequence from Namulonge in south-central Uganda (KJ606250) with which they shared 97.3% homology. Previously unpublished sequences described in this study were deposited in GenBank with the accession codes (MF511057 to MF511060). Sequences presented in Figure 7 with their GenBank codes are as follows: MF511057 (Uvira553), MF511058 (Rutshuru586), MF511059 (Rutshuru27), MF511060 (Rutshuru9), MF511061 (Rutshuru4), MF511062 (Ituri574), MF511063 (Ituri573), MF511064 (Ituri572), MF511065 (Ituri568), MF511066 (Ituri562u), MF511067 (Ituri562c), and MF511068 (Ituri557).

Download as PowerPoint

Discussion

CBSD is a growing problem affecting cassava production in East, Southern, and Central Africa. Although it was confined to coastal eastern Africa for several decades following its first report from coastal Tanzania (Storey 1936), since 2004 it has spread widely through midaltitude zones of East and Central Africa (Alicai et al. 2007; Legg et al. 2011, 2015). The first confirmation of the occurrence of CBSD in DRC, supported by molecular diagnostic tests, was made from cassava leaf samples collected in Lubero Territory, Nord-Kivu Province (Mulimbi et al. 2012). Our study reports the results of the first comprehensive regionwide survey of CBSD in eastern DRC. Monitoring the pattern of development of CBSD in this part of Africa is particularly critical, since cassava is the most important staple food crop in Central Africa, including DRC. Additionally, there is a high risk that CBSD will spread from Central to West Africa through DRC. DRC straddles the center of the continent, and spread of CBSD from the east to the west of the country would pose a major threat to the important cassava-producing countries of West Africa, notably Nigeria, which is the world’s largest cassava producer.

Symptoms of CBSD observed during the surveys that ran from late 2015 to 2018 were typical of those associated with the disease in other parts of East and Central Africa and as originally described from Tanzania by Nichols (1950). Symptoms resembling those caused by CBSD have been reported widely from different parts of Central and West Africa, including western DRC (Mahungu et al. 2003). However, in all such cases, symptoms have been confined to the tuberous roots, and characteristic foliar and stem symptoms have not been described. In the study reported here, all of the three main foliar symptoms (veinal chlorosis or chlorotic blotching on lower leaves, necrotic lesions on green stems, and stem dieback in severe infections) were observed in all territories where the presence of CBSD was also confirmed through virus diagnostic tests. New reports made here confirm the expansion of the area known to be affected by CBSD to two additional provinces (Sud-Kivu and Haut-Katanga) and nine additional territories. Administrative zones reported as affected for the first time cover a total area of >62,000 km². Additionally, affected areas of Pweto Territory in southeastern DRC are >900 km away from the locations of previously published reports of CBSD in the country. As such, the prevalence of CBSD reported here represents a major expansion of the CBSD pandemic in Africa. It is significant that Pweto Territory shares a border with northern Zambia, another country/region that has reported a recent outbreak of CBSD (Mulenga et al. 2018). The surveys reported here represent the first geographically extensive record of CBSD in DRC. As such, the data obtained provide an important baseline against which future assessments of the disease and its epidemiology can be measured. They also provide an important complement to country or regionwide assessments of CBSD, including Tanzania (Jeremiah et al. 2015; Legg and Raya 1998), Mozambique (Hillocks et al. 2002), Rwanda (Munganyinka et al. 2018), and Uganda (Alicai et al. 2019), which contribute to building a continental picture of the development of the CBSD pandemic.

The lower severity of CBSD symptoms in Haut-Katanga Province may be a consequence of the more recent infection of many of the fields in this region, as evidenced by the large increase in incidence from 2016 to 2018. CBSD is known to become more severe over repeated cropping cycles, resulting in “degeneration” of cassava varieties (Shirima et al. 2019). It is also possible, however, that this effect is related to varietal interactions or virus strain variation (Mohammed et al. 2012). Varieties differ greatly in their responses to CBSIs (Kaweesi et al. 2014; Shirima et al. 2019), and patterns of distribution of varieties can influence CBSD incidence and severity in the regions in which they occur (Ndyetabula et al. 2016). Also, strains of CBSIs are known to vary in the severity of symptoms that they induce (Mohammed et al. 2012) and Haut-Katanga was unique as a province in which only CBSV was detected. Further investigation would be required to determine which of these possible causes are responsible for the symptom severity differences observed.

In common with most important cassava-growing regions of sub-Saharan Africa, there was a high level of varietal diversity in the eastern provinces of DRC, although about a quarter of these accounted for two-thirds of the sampled fields. Both local and improved varieties in farmers’ inspected fields were infected and susceptible to CBSIs. Two varieties (Mwabayilo and Nyamwalirwa) had no symptoms of infection even though they were widely cultivated in Sud-Kivu where CBSD was otherwise widespread. This suggests that they may be relatively more resistant to CBSD than other varieties being grown in the same area, although varietal comparison experiments planted at a single location with known inoculum pressure conditions would be required to verify this possibility. An important practical implication of the increased geographical coverage and incidence of CBSD is that cassava varieties possessing both whitefly and disease resistances are urgently needed. Previous initiatives to facilitate the introduction and mass dissemination of CMD-resistant varieties in eastern DRC were undertaken as part of the Great Lakes Cassava Initiative, which ran from 2007 to 2012 (Catholic Relief Services 2012). More than 17 million stems of these varieties were distributed in Burundi, DRC, Kenya, Rwanda, Tanzania, and Uganda as planting material during this time, reaching >800,000 farmer beneficiaries. Six of the 10 improved CMD-resistant varieties recorded from the current survey were distributed during this initiative, including Sawasawa (MM96/3920), Sukisa (MM96/1666), Liyayi (MM96/0287), Mayombe (MM96/7752), and Nabana (MM96/4653), and one more (Obama; TME 419) was evaluated in participatory varietal selection trials. These seven were the primary varieties being cultivated at 49 of the surveyed sites, equivalent to 12.6% of the total, and one or more were present in all of the provinces. Although these have clearly had a significant impact in sustaining cassava production in eastern DRC over the last decade, it has been widely recognized that varieties resistant to CMD are mostly susceptible to CBSD (Alicai et al. 2007; Legg et al. 2006). This phenomenon was supported by our data, although there was no evidence to suggest that the improved CMD-resistant varieties were more or less susceptible to CBSD than local varieties; the two varieties with the highest CBSD incidences and most severe symptoms were both local varieties. Although there were several widely cultivated varieties without any incidence of CBSD (Supplementary Table S1), it is not possible to conclude from these survey results that any of these had resistance to CBSD. Controlled experiments would be required to verify this possibility. Studies of this type have been conducted in coastal Tanzania and used to identify several local varieties with promising levels of resistance to CBSD (Masinde et al. 2018). Most of the current cassava improvement focus in Africa, however, is aimed at combining sources of resistance to both CMD and CBSD (Kawuki et al. 2016) as well as using transgenic approaches to engineer CBSD resistance into varieties that already carry resistance to CMD (Wagaba et al. 2017).

Although the overall abundance of B. tabaci was moderate to low during the survey periods in 2016 and 2018, 15% of fields had “abundant” whitefly populations (sensu Legg et al. 2011), and these fields occurred throughout the surveyed area. In addition, a substantial proportion (55%) of the sampling was conducted during cool dry seasons (May to September), which may have contributed to the overall moderate to low whitefly abundances: cassava-colonizing B. tabaci is known to thrive during periods of hot, moist weather (Dengel 1981; Fargette et al. 1993). Increased abundance of B. tabaci whiteflies has been widely recognized as a major factor driving CBSD spread in East and Central Africa (Hillocks et al. 1999; Legg et al. 2011).

UCBSV as the causal virus of CBSD was first reported from two territories of Nord-Kivu in eastern DRC from surveys conducted in late 2011 (Mulimbi et al. 2012). CBSD was clearly very rare at that time, as UCBSV was only detected from two out of 100 farmer fields surveyed in Beni and Lubero territories. In 2019, the occurrence of both CBSV and UCBSV was reported further north from Ituri Province (Casinga et al. 2019). Results presented in our study greatly expand the known distribution of CBSIs in DRC. This includes the new reporting of UCBSV from Sud-Kivu, the expanded presence of UCBSV in Nord-Kivu (including Rutshuru, Masisi, and Nyiragongo territories), as well as the new occurrence of CBSIs in Haut-Katanga Province (CBSV). This information on patterns of occurrence of these two viruses reflects the general situation elsewhere from Kenya, Tanzania, and Uganda, in which both species are reported from most regions affected by CBSD. Both viruses are reported from Kenya (Adams et al. 2013), Malawi (Mbewe et al. 2015), Tanzania (Mbanzibwa et al. 2011a), and Uganda (Kaweesi et al. 2014), although exceptions to this pattern are Mozambique, from which only CBSV has been detected (Amisse et al. 2019; Mohammed et al. 2012), and Burundi, in which only UCBSV has been identified (Bigirimana et al. 2011). However, given the occurrence of CBSV in countries surrounding Burundi, including Rwanda, Tanzania, and DRC, it seems highly likely that CBSV will also spread there in the near future.

Sequence data revealed that UCBSV sequences from eastern DRC were most closely related to UCBSV isolates from central, southern Uganda and that there was a high degree of homology within the DRC sequences themselves. The generally high level of homology among UCBSV sequences from diverse locations throughout East and Central Africa, however, means that it cannot be assumed that these isolates, now in DRC, originated directly from Uganda. However, the cryptic nature of CBSD symptoms means that infected planting material can be inadvertently moved from place to place, and there have been both unofficial and official large-scale movements of cassava stems between countries in East and Central Africa that may have unintentionally resulted in the spread of CBSIs within the region. CBSV sequences were more divergent from nearest relatives in the GenBank database. These findings are unsurprising, as an analysis of a large group of sequences from East and Southern Africa revealed a generally higher level of variability among CBSV isolates than UCBSV (Mbanzibwa et al. 2011b). Future studies will be required, however, to examine a larger set of both UCBSV and CBSV isolates from eastern and southeastern DRC in order to improve the resolution in the understanding of CBSI diversity in DRC.

The CBSD surveillance study described here has provided an important update on the status of this highly damaging disease of cassava. Evidence was obtained for an increase in the occurrence of the disease from 2016 to 2018, particularly in the southern part of the sampling domain, in Haut-Katanga. Central DRC is a sparsely populated forest, which will impede the westward spread of the disease. However, regions to the west of Haut-Katanga, including Haut-Lomami and Lualaba, are important savanna-based cassava-growing regions and may provide a route through which CBSD can be carried further westward. CBSD management strategies for East Africa have comprised both the development/deployment of resistant varieties (Jennings 1960; Kawuki et al. 2016) and the application of phytosanitary measures (Hillocks and Jennings 2003; Legg et al. 2017). The implementation of such strategies is urgently required for the provinces of eastern DRC currently affected. Arguably more important for the mitigation of the CBSD pandemic, however, will be the prevention of spread to provinces further to the west, as this increases the threat that the disease poses to important production centers feeding the metropolis of Kinshasa as well as other key cassava-producing zones in Central/West Africa. To achieve this, government institutions, nongovernmental organizations, and international partners will need to put in place systems to raise awareness of the threat posed by CBSD, monitor its spread, prevent movements of planting material from east to west, and identify, multiply, and disseminate CBSD-resistant varieties.

Acknowledgments

The authors gratefully acknowledge the substantial assistance received from farmers, researchers, and agricultural staff in DRC. The authors also thank Dr. Everlyne Wosula of IITA-Tanzania for assistance with some of the statistical analyses.