Life history shifts in an exploited African fish following invasion by a castrating parasite

Abstract Evolutionary theory predicts that infection by a parasite that reduces future host survival or fecundity should select for increased investment in current reproduction. In this study, we use the cestode Ligula intestinalis and its intermediate fish host Engraulicypris sardella in Wissman Bay, Lake Nyasa (Tanzania), as a model system. Using data about infection of E. sardella fish hosts by L. intestinalis collected for a period of 10 years, we explored whether parasite infection affects the fecundity of the fish host E. sardella, and whether host reproductive investment has increased at the expense of somatic growth. We found that L. intestinalis had a strong negative effect on the fecundity of its intermediate fish host. For the noninfected fish, we observed an increase in relative gonadal weight at maturity over the study period, while size at maturity decreased. These findings suggest that the life history of E. sardella has been shifting toward earlier reproduction. Further studies are warranted to assess whether these changes reflect plastic or evolutionary responses. We also discuss the interaction between parasite and fishery‐mediated selection as a possible explanation for the decline of E. sardella stock in the lake.

For fish, both natural predation and fishing (i.e., predation by humans) are important selective factors that drive adaptive changes in life history traits such as developmental rates and timing of reproduction (Heino & Godø, 2002;Jorgensen et al., 2007;Jørgensen et al., 2009;Sharpe et al., 2012). Fishing practices and predation are usually nonrandom factors, as gears are often designed to selectively take larger and older fish in the population (Law, 2000). In this case, smaller fish are likely to have a higher probability of survival than the larger ones, and among them, those that can mature and reproduce early will be selected (Jorgensen et al., 2007;Jørgensen et al., 2009). Assuming that early maturation is heritable to some extent, this should result in life histories changing toward earlier reproduction at smaller sizes (Ayllon et al., 2015;Heath et al., 2002;Olsen et al., 2004;Sinclair-Waters et al., 2020).
Increased reproductive effort in hosts exposed to castrating parasites has been reported in a number of species. So far, however, most documented life history changes seem to result from adaptive plastic responses of hosts to parasitic exposure, more than life history evolution following a change in parasite-mediated selection (Chadwick & Little, 2005;Hudson et al., 2019;Vale & Little, 2012).
In this study, we investigated whether the castrating parasitic cestode Ligula intestinalis was responsible for a life history change in the cyprinid fish Engraulicypris sardella in Lake Nyasa. We studied the freshwater fish E. sardella, which is the second intermediate host for the cestode L. intestinalis. E. sardella (Günther, 1868), locally known as Usipa or Lake Malawi sardine, is a small, slender, silvery, zooplanktivorous fish endemic to Lake Nyasa (Lowe-McConnell, 1993;Rufli & Van Lissa, 1982) that occurs in shoals, which are widely distributed within the lake and found in both nearshore areas and offshore pelagic water, down to a depth of approximately 200 m (Maguza-Tembo et al., 2009).
Engraulicypris sardella is an annual species, where hatchlings grow and age to reproduce and die in a yearly cycle (Iles, 1960), although some studies indicate that they can live longer (Rusuwa et al., 2014;Thompson & Bulirani, 1993). They have been reported to breed throughout the year but with bi-annual recruitment peaks occurring during the wet season and dry season (Morioka & Kaunda, 2005;Rusuwa et al., 2014).
During early developmental stages, E. sardella feeds exclusively on phytoplankton and then switches to feeding on zooplankton upon reaching adulthood Degnbol, 1982). E. sardella demonstrates a rapid growth rate and can attain a maximum total length of about 130 mm in a year (Thompson, 1996;Tweddle & Lewis, 1990). Males and females mature at a size of about 70 and 75 mm, respectively (Thompson & Allison, 1997;Thompson et al., 1996).
Engraulicypris sardella forms an important part of the food web of Lake Nyasa. The species is primary consumer of zooplankton (Degnbol, 1982;Konings, 1990) and an important prey for pelagic piscivorous fishes, particularly Diplotaxodon spp. and Rhamphochromis spp. , as well as piscivorous birds (Linn & Campbell, 1992). E. sardella is also of high commercial value, and for many decades, it has been the main animal protein source for most of the local human population (Manyungwa-Pasani et al., 2017).
However, recently it has been observed that these cyprinids are infected by the cestode L. intestinalis.
Ligula intestinalis (L. 1758) is a common and widespread cestode, that uses cyprinid fish as the second intermediate host (Dubinina, 1980;Kennedy, 1974). The parasite is trophically transmitted and has a complex life cycle involving two aquatic intermediate hosts, a planktonic copepod and a fish (Dubinina, 1980;Loot et al., 2001). It reaches sexual maturity in the abdominal cavity of piscivorous birds that are the final hosts (i.e., the hosts where parasite reproduction takes place) (Dubinina, 1980;Loot et al., 2001).
In infected fish, the parasite is found filling the body cavity (Hoole et al., 2010). Higher infection rates are observed in larger and older E. sardella than in juvenile individuals (Msafiri et al., 2014;Rusuwa et al., 2014), which can partly be explained by diet shifts from phytoplankton to zooplankton as E. sardella reaches maturity.
The invasion of L. intestinalis in Lake Nyasa was first noted in the late 1990s during longline research surveys where a milkish white worm was found in the body cavity of the endemic pelagic cyprinid fish E. Sardella (Mwambungu et al., 1996). The worm was identified to be the tapeworm Ligula intestinalis (L.). This parasite is believed to be introduced into Lake Nyasa by migrating fish-eating birds such as the White-breasted cormorant (Phalacrocorax carbo), which is one of the most abundant fish-eating birds in the Lake Nyasa basin (Linn & Campbell, 1992) and one of the final hosts of L. intestinalis (Loot et al., 2001;Rosen, 1920). In Lake Nyasa, this cestode has been increasingly reported since it was first noted by Mwambungu et al. (1996). E. Sardella appears to be the only species used as intermediate fish host (Gabagambi et al., 2019;Gabagambi & Skorping, 2018;Msafiri et al., 2014;Rusuwa et al., 2014) ( Figure   S1).
Ligula intestinalis is known to induce castration in several intermediate hosts (Cowx et al., 2008;Hoole et al., 2010;Kennedy et al., 2001;Loot et al., 2002;Wyatt & Kennedy, 1988) and has therefore been suggested to cause population crashes of its host (Burrough et al., 1979;Kennedy et al., 2001). This could sometimes lead to local extinction of the parasite in small ecosystems (Kennedy et al., 2001). Recent results, however, indicate that local extinction of this parasite is unlikely in Lake Nyasa due to spatial and temporal variations in transmission rates (Gabagambi & Skorping, 2018).
Under such conditions of recent invasion, we hypothesize that the cestode L. intestinalis should select for a shift in resource investment from somatic growth toward reproduction in its intermediate fish host E. sardella. Using data collected from 2005 to 2015 in the northern part of Lake Nyasa, we address the following three questions: (i) What are the effects of L. intestinalis on the fecundity of E. sardella? (ii) has reproductive investment at maturity of E. sardella increased over time? and (iii) has the average size at maturity of E. sardella decreased?
We then further discuss the selective roles of parasitic invasion versus other environmental factors that may recently have changed in Lake Nyasa.

| Study area
The study was conducted in the northern part of Lake Nyasa, Tanzania side ( Figure 1). Lake Nyasa, also known as Lake Malawi in Malawi and Lago Niassa in Mozambique, is the southernmost great lake in the East African Rift Valley system, located between Malawi, Mozambique, and Tanzania. The lake is the third largest freshwater lake in Africa after lakes Victoria and Tanganyika and is the second largest lake by volume after Lake Tanganyika (Darwall et al., 2010;Hampton et al., 2018;Macuiane et al., 2015). The lake has a maximum depth of 785 m, a volume of 8400 km 3 , a surface area of 29,000 km 2 , approximate length of 550 km, and mean width of around 48-60 km and is located 472 m above the sea level (Bootsma & Hecky, 1993;Darwall et al., 2010;Gonfiantini et al., 1979;Patterson & Kachinjika, 1995). The total catchment area of the lake is 126,500 km 2 (Kumambala & Ervine, 2010) of which 97,750 km 2 is land catchment (Menz, 1995). The mean surface temperature of the lake is between 24 and 28°C (Vollmer et al., 2005) and the annual rainfall ranges between 1000 and 2800 mm (LNBWB, 2013). The lake experiences two main seasons, the dry season (May-August) and wet season (November-April), which are governed by the regional climate (Lyons et al., 2011;Vollmer et al., 2005). Lake Nyasa is meromictic, although it may experience mixing during the dry season in the southern tip of the lake where the depth is relatively shallow (Darwall et al., 2010;Vollmer et al., 2005;Weyl et al., 2010). Due to the stratification, together with the great depth of the lake, the nutrient availability to the plankton community are very low, and thus, the lake is considered "oligotrophic" (Irvine et al., 2001;Mwambungu & Ngatunga, 2001). The lake has more than 1,000 different fish species, many of which are endemic (Chafota et al., 2005;Salzburger et al., 2014). Sampling was conducted at Wissman Bay that is located at the northern end of the lake (sampling stations of Matema S9°29ʹ; E34°01ʹ, Mwaya S9°33ʹ; E33°57ʹ, Kafyofyo S9°35ʹ; E33°57ʹ, and Kiwira S9°37ʹ; E33°57ʹ). The fishing procedure involved nine crew members using two dugout canoes and one large plank boat. On the fishing ground, one of the dugout canoes was equipped with pressurized paraffin lamps (between one and three) and was stationed with one crew member away from the remaining vessels. The artificial light was used to concentrate the fish into the given area. This process took several hours.

| Sampling procedure
After a sufficient amount of fish had been attracted, the other unlit fishing vessels simultaneously deployed a net in a semicircular shape around the concentrated fish, and this was hauled by hand into the plank boat. A total of 3488 female E. sardella were sampled (Table 1).
Males were also caught and examined as part of general monitoring, but due to the low reliability of stage determination for males of such a small fish species, only females were included in this study.
Upon landing, the total length and weight of each E. sardella were measured to the nearest 5 mm and 0.01 g, respectively. Specimens of E. Sardella were kept in cool boxes until further examination. The fish were later dissected for parasite determination. L. intestinalis was identified according to the protocol by Dobben (1952) while examination of other parasites was done according to Parpena (1996).
The sex of E. sardella was determined using a stereomicroscope (Wild Heerbrugg M5) at 6.4X magnification. Gonad maturity was assessed on a seven-stage maturity scale (Table 2), modified from Holden and Raitt (1974).
Therefore, we were able to maintain a good level of consistency and accuracy in the determination of maturity stage across our sampling period. In , 2008  Gonads were weighed to the nearest 0.01 g (wet weight) using sensitive precision balances (vwr™-model ECN 611-2315 and Endel™-model WPS) and fecundity for infected and noninfected female E. sardella was determined through gravimetric methods (Holden & Raitt, 1974) by counting the advanced yolked oocytes present in ripe and gravid E. sardella. The complete ovary was taken out and preserved in modified Gilson's fluid (100 ml 60% alcohol, 800 ml water, 15 ml 80% nitric acid, 18 ml glacial acetic acid, 20 g mercuric chloride) for 24 hr. Thereafter, the ovaries were shaken periodically to help loosen the eggs from connecting ovarian tissues.
After the eggs were liberated from the ovarian tissues, they were washed thoroughly, spread on blotting paper, and allowed to dry at ambient temperature ranging between 25 and 30°C. Thereafter, the total numbers of eggs were weighed to the nearest 0.01 g using sensitive precision balance to have a total weight of eggs. Afterward, we collected a random subsample of the eggs, which were weighed and counted out on petri dish subsections using a stereomicroscope (Wild Heerbrugg M5) at 6.4× magnification. The total number of eggs (i.e., fecundity) in the ovaries was calculated following the formula given by Holden and Raitt (1974) as follows: F = nG/g, where n = number of eggs in subsample, G = total weight of eggs from the ovary, and g = weight of the subsample. Fish somatic weight was determined by subtracting the gonad weight from the total weight of the fish.

| Statistical analyses
All statistics and graphics were carried out using R, version 3.2.5 (http://r-proje ct.org). 1: mature) and body length as a continuous predictor variable ( Figure S2). From the parameters of these logistic regression equations, and following Diaz Pauli and Heino (2013), we estimated for each year the length at which the probability of maturing is 50% (i.e., LM 50 ):

TA B L E 1
where p is the probability of maturity (0.5), a is the intercept, and b is the slope.
To test whether LM 50 decreased over time, we fitted a linear model (lm) with LM 50 as a response variable and year as a numerical predictor (linear and quadratic terms).
Reproductive investment at maturity (relative weight of gonads at stage IV) in noninfected E. sardella increased significantly from 2005 to 2015 (glm, estimate = 0.14 ± 0.01, df = 1, t = 9.59, p < .001; Figure 3).  Figure 4).  Year Relative gonad weight at maturity (%) the species range of this parasite (Barson & Marshall, 2003;Carter et al., 2005;Cowx et al., 2008). We also found that the relative weight of gonads increased, while body size at maturity decreased, over the 10-year duration of this study. These temporal changes, found in noninfected fish, indicate that investment of E. sardella into early reproduction has increased at the expense of somatic growth.

| D ISCUSS I ON
This study took place a few years only after the arrival of L. intestinalis in the lake. A parasitic relationship between L. intestinalis and E. sardella in Lake Nyasa was indeed first observed in 1996 (Mwambungu et al., 1996). An earlier study investigating the breeding biology and in particular examining the ovaries of E. sardella between 1992 and 1994, did not report any case of L. intestinalis infection (Thompson, 1996). This tapeworm was thus likely absent from Lake Nyasa prior to the late 1990s. After the first observation, E. sardella in the lake kept being found infected by L. intestinalis, as manifested by the work of J.K. Kihedu (MSc thesis, Sokoine University of Agriculture, Tanzania, 2006, unpublished data). The earliest sampling year in our study is 2005, when prevalence is estimated at 50% (Table 1). This indicates that L. intestinalis had spread and therefore that the selection caused by this parasite on its host had increased steadily during the early years after introduction. Our study remains correlative, yet given the timing of the observed life history shift relative to the invasion of the lake by L. intestinalis, it seems legitimate to consider parasitism as a likely contributing factor.
In general, changes in age-specific mortality or fecundity rates lead to changes in selection on life history traits. In our study, we observed an overall 69% lower fecundity in infected versus uninfected hosts, that is, the cestode L. intestinalis caused a significant partial castration in E. sardella. Reduced host fecundity is a common outcome of parasite infection (Gooderham & Schulte-Hostedde, 2011;Hurd, 2001), but is especially severe for castrating parasites. Castration selects for higher, earlier reproductive effort, as those individuals that are able to reproduce before castration are clearly favored (Forbes, 1993). A number of host species have been shown to increase their early reproductive effort when parasitism reduces their chances for future reproduction (Adamo, 1999;Jokela & Lively, 1995;Lafferty, 1993b;Minchella & Loverde, 1981). This kind of adaptive response can result from two distinct mechanisms, namely plasticity or evolution, and distinguishing between the two can reveal challenging.
Plastic life history shifts toward increased investment in early reproduction in exposed and/or infected hosts have been reported for a range of host-parasite systems. In insects, Polak and Starmer (1998) observed that experimentally parasitized male Drosophila nigrospiracula infected with a mite (Macrocheles subbadius) lived shorter lives, but before dying they courted females significantly more than nonparasitized controls. Further, Adamo (1999) observed that female crickets (Acheta domesticus) increased egg laying in response to infection with the bacterium Serratia marcescens. In snails, Minchella and Loverde (1981) and Thornhill et al. (1986) observed an increase in reproductive output in female Biophalaria glabrata parasitized by a castrating trematode Schistosoma mansoni. In crustaceans, Chadwick and Little (2005)  In reptiles, Sorci et al. (1996)  Year LM50 for uninfected fish (mm) is found in noninfected hosts and therefore cannot be explained by plastic responses to infection. In addition, given the empirical evidence available at this stage, plastic responses to exposure appear unlikely, given the lack of clear correlation between yearly fluctuations in prevalence and life history trends, as one would expect under such a scenario. We therefore cannot exclude that our results may reflect adaptation to recent changes in Lake Nyasa.
Importantly, increased parasite pressure may not be the only environmental change that has taken place in Lake Nyasa over the last couple of decades and that might have triggered life history responses in E. sardella. Other potential sources of selection for earlier reproduction include fishing (Fenberg & Roy, 2008;Heino & Godø, 2002;Hutchings & Fraser, 2008;Jorgensen et al., 2007;Jørgensen et al., 2009;Kuparinen & Merilä, 2007;Sharpe & Hendry, 2009;Sharpe et al., 2012); increased predation by native or introduced species (Hampton et al., 2018;Sharpe et al., 2012); and fluctuations in zooplankton abundance that may induce earlier maturation.
Most evidence of fishery-induced evolution comes from large, heavily exploited fish population stocks (e.g., North Arctic cod) where industrial fishing using trawlers has been in practice for many years. On the contrary, the Lake Nyasa E. sardella fishery is mainly traditional, operating in nearshore lake zones using paddled dugout canoe crafts (Mwambungu & Ngatunga, 2001). In the last years of this study, however, E. sardella stocks have collapsed, despite no sudden changes in fishing effort. As a consequence fishing pressure has dramatically increased in Wissman bay ( Figure   S3).
In the present study, E. sardella were sampled using the traditional fishing method. The majority of the sampled fish was composed of individuals of the body sizes between 50 and 100 mm in length, which corresponds to mature fish (i.e., from stage IV and above). This suggests that the traditional E. sardella fishing practice is probably size-selective and induces a higher mortality in adults than younger fish, thus possibly reinforcing the selective effects of parasitism. Interestingly, the dramatic decrease in landings in 2013 was preceded by three consecutive years with high L. intestinalis prevalence ( Figure S3), further suggesting that parasitism is a strong selective factor. In this system, L. intestinalis may have acted synergistically with fishery-mediated selection in driving what appears like an evolutionary shift toward earlier reproduction of E. sardella in Lake Nyasa.
Increased predation by native or introduced organisms could also be one factor affecting selection on life history traits of E. sardella. In the native cyprinid fish Rastrineobola argentea in Napoleon Gulf of Lake Victoria, Sharpe et al. (2012) observed decreased body size, maturation at smaller sizes, and increased reproductive effort in response to the introduced predator fish Lates niloticus.
However, in contrast to Lake Victoria and many other ancients lakes where dozens of non-native species have been introduced over the past decade (Hampton et al., 2018), in Lake Nyasa no new introduced predator for E. sardella has been reported so far. The primary natural piscivorous predators of E. sardella in this lake are the pelagic haplochromine cichlids from the genera Ramphochromis, Unfortunately, the area where the present study was conducted is a data-poor region; the last pelagic ecosystem stock assessment was conducted between 1991 and 1994 (Menz (1995). Recent time series on abundance fluctuations of the natural predators of E. sardella are lacking. Further research, particularly on the combined effects of parasitism, fishing, and natural predation on E. sardella in Lake Nyasa, would be highly valuable, given the ecological and economical importance of this fish species.
Another factor that could have affected selection on the life history traits of E. sardella in Lake Nyasa may be parallel increases in the prevalence of other parasites. In their natural habitats, hosts are usually infected by two or more different parasite species (Kotob et al., 2017;Petney & Andrews, 1998). To the best of our knowledge, the only other parasite that has been reported to infect E. sardella is the nematode Camallanus sp. (Mgwede & Msiska, 2018). In the present study, we caught 3,488 wild, that is, naturally infected E. sardella, none of them observed with Camallanus sp. infection.
Overall, this study reveals that life history of E. sardella in Lake Nyasa has been shifting, over a period corresponding to the invasion of this lake by a castrating parasite. It is correlative, and the causative links between parasitism and life history changes remain to be established. Yet, the cestode L. intestinalis, by strongly reducing the fecundity of its host, appears as a likely driver of life history evolution, similar in its effects to size-selective fisheries. In Lake Nyasa, these two types of selective factors may have acted concomitantly. More work is now warranted to examine the origin of these changes and determine whether they represent plastic or evolutionary responses.

ACK N OWLED G M ENTS
The data presented in this manuscript were collected by research scientists from the Tanzania Fisheries Research Institute (TAFIRI), Kyela Centre, together with the large number of students from the University of Dar es Salaam (UDSM) and Sokoine University of Agriculture (SUA). We would like to thank all of them for sharing their data with us. We also thank Dr. Madeleine Carruthers for smoothing the English writing.

CO N FLI C T O F I NTE R E S T
None declared.

E TH I C S S TATEM ENT
This research received ethical approval from Tanzania Fisheries Research Institute (Application ID: TAFIRI/HQ/PF637/100).

O PE N R E S E A RCH BA D G E S
This article has earned an Open Data Badge for making publicly available the digitally-shareable data necessary to reproduce the reported results. The data is available at https://doi.org/10.5061/ dryad.p2ngf 1vp3.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sets supporting this manuscript can be accessed through։ https://doi.org/10.5061/dryad.p2ngf 1vp3.