Received: 1 February 2019 | Revised: 26 June 2019 | Accepted: 27 June 2019 DOI: 10.1111/fwb.13383 ORIGINAL ARTICLE Long‐term change of phytoplankton in Lake Kivu: The rise of the greens Edouard Rugema1 | François Darchambeau2 | Hugo Sarmento3 | Maya Stoyneva‐Gärtner4 | Maria Leitao5 | Wim Thiery6 | Adrien Latli7 | Jean‐Pierre Descy2 1Symbion Power Lake Kivu Ltd, RSSB Headquarters, Kigali, Rwanda Abstract 2Unit of Chemical Oceanography, University 1. Phytoplankton time‐series enable the observation of recurrent seasonal patterns of Liège, Liège, Belgium on community composition, but drastic shifts in community structure are rarely 3Departamento de Hidrobiologia (DHb), Universidade Federal de São Carlos, observed, unless linked to anthropogenic impacts (e.g. changes in nutrient inputs). São Carlos, Brazil Here, we report a unique case of non‐seasonal, multiannual regime shift of phyto‐ 4 Department of Botany, Faculty of plankton community structure in Lake Kivu, East Africa, and explore the possible Biology, Sofia University ‘St Kliment Ohridski’, Sofia, Bulgaria causes for such changes. 5Bureau d’études Bi‐Eau, Angers, France 2. We recorded phytoplankton biomass every 2 weeks over a period of 12 years 6 Department of Hydrology and Hydraulic (from 2002 to 2015). Phytoplankton community previously dominated by cyano‐ Engineering, Vrije Universiteit Brussel, Brussels, Belgium bacteria and diatoms changed considerably from 2012, with a conspicuous rise 7DEMNA, Service Public de Wallonie, in the biomass of green algae (from 3 to 33% lake‐wide), and a decline of crypto‐ Gembloux, Belgium phytes and cyanobacteria along with an increase in total chlorophyll a concentra‐ Correspondence tion. Concomitantly, we observed significant differences in the mixed layer depth Descy Jean‐Pierre, Unit of Chemical between the two periods, resulting from changes in the thermal gradient. The Oceanography, University of Liège, Liège, Belgium. euphotic zone depth was also reduced in 2012–2015, probably from the higher Email: [email protected] phytoplankton biomass. Moreover, mean particulate nitrogen and phosphorus in‐ Funding information creased as well as the particulate N:P ratio. Conselho Nacional de Desenvolvimento 3. Desmids (Cosmarium laeve) now dominate the mixolimnion, probably due to atelo‐ Científico e Tecnológico, Grant/Award Number: 309514/2017‐7; Fundação mixis, observed in high‐frequency temperature measurements. We believe that Amazônia Paraense de Amparo à Pesquisa, decadal or multidecadal variability of climate on the Indian Ocean may be the Grant/Award Number: 2014/14139‐3 reason for this regime shift in thermal stratification, with a strong impact in phy‐ toplankton community structure. 4. In conclusion, our study demonstrates that tropical lakes may be subject to non‐ seasonal fluctuations that generate multiannual regime shifts. These regime shifts have probably never been reported previously because of the lack of long phyto‐ plankton time‐series in low latitudes. KEYWORDS atelomixis, climate variability, community structure, East Africa, tropical great lake Freshwater Biology. 2019;00:1–16. wileyonlinelibrary.com/journal/fwb © 2019 John Wiley & Sons Ltd. | 1 2 | RUGEMA ET AL. 1 | INTRODUCTION nutrient limitation in the rainy season and low light and high nutri‐ ents during the dry season has been documented in classic studies As phytoplankton can respond quickly to changes in tempera‐ in African lakes (e.g. Hecky & Kling, 1987; Kilham, Kilham, & Hecky, ture, light, nutrient loading, and water column physical structure, 1986) and is a key to understanding shifts in phytoplankton compo‐ the composition of the assemblage—both taxonomical and func‐ sition in these lakes. Resource ratios may also play a role in determin‐ tional—may reveal natural and man‐induced changes in these factors ing which species may be the most successful in given conditions of (Reynolds, 2006). There are well‐documented examples of long‐term light and nutrient availability (Kilham et al., 1986). monitoring of phytoplankton in temperate lakes, particularly for as‐ Although it has not been used extensively so far, pigment‐based sessing effects of lake acidification (Ginn, Cumming, & Smol, 2007; analysis of phytoplankton composition may provide an adequate Vrba et al., 2003), eutrophication or oligotrophication (e.g. Anneville, framework to relate community composition to environmental Ginot, Druart, & Angeli, 2002; Ruggiu, Morabito, Panzani, & changes. Phytoplankton marker pigments, determined by high‐per‐ Pugnetti, 1998), or global warming (e.g. Sarmento, Amado, & Descy, formance liquid chromatography (HPLC) analysis, have been used for 2013; Stenuite et al., 2007). The response at the level of individual assessing biomass at the class level, with many applications in marine, cell populations can be fast (days to weeks) and reflect environmen‐ estuarine, and freshwater environments (e.g. review in Sarmento & tal or seasonal variations, but changes in assemblage composition Descy, 2008). Estimating algal abundance from marker pigment re‐ resulting from a change in lake status (e.g. from a reduction of nutri‐ lies on calculations using measured pigment concentrations and ra‐ ent loading) occur at longer time scales (Reynolds, 1992). Therefore, tios of those markers to Chla (e.g. with CHEMTAX, Mackey, Mackey, years of continuous observations are necessary to tell apart changes Higgins, & Wright, 1996). As it is based on a fast, automatic and re‐ related to significant environmental trends from stochastic or sea‐ liable analytical technique, the pigment approach has been largely sonal variations. Typically, decades of phytoplankton monitoring, used in oceanographic studies and monitoring programmes (Jeffrey, using the same methods, are needed to assess those trends. Mantoura, & Bjørnland, 1997), as well as in large lake studies (Descy In tropical lakes, few studies based on phytoplankton surveys et al., 2005; Fietz & Nicklisch, 2004; Fietz, Sturm, & Nicklisch, 2005; have been long enough to assess changes at a multi‐annual scale. Sarmento et al., 2006). We carried out a long‐term phytoplankton Most of the time, long‐term surveys were based on measurements monitoring of the water column of Lake Kivu using this approach. of chlorophyll a (Chla) concentration (i.e. an estimate of phytoplank‐ Lake Kivu is a great and deep (maximum depth of 489 m) ton biomass), as a proxy for pelagic primary production, and gaps are meromictic lake of the East African Rift (Figure 1). It is located north often present in time series. For instance, records of Chla in Lake of Lake Tanganyika, at 1463 m above sea level. In this lake, the mixo‐ Tanganyika (Bergamino et al., 2010; Horion et al., 2010) have shown limnion (i.e. the mixed layer of a meromictic lake) alternates between variations at seasonal and inter‐annual scales, related to variations periods of complete mixing down to maximum 65 m and periods of of the mixed layer depending on meteorological and climatic con‐ stratification, during which nutrients become depleted in the eu‐ ditions, which determine water column processes in African Great photic zone (Figure 2; Sarmento et al., 2006; Schmid & Wüest, 2012; Lakes (Nicholson, 1996; O'Reilly, Alin, Plisnier, Cohen, & McKee, Thiery, Martynov, et al., 2014; Thiery, Stepanenko, et al., 2014). The 2003; MacIntyre, 2012; Tierney et al., 2010; Tierney, Smerdon, deepening of the mixed layer occurs during the dry season (June, Anchukaitis, & Seager, 2013). However, the lack of continuity in July, August) and is primarily driven by the substantial decrease in records (Plisnier, Nshombo, Mgana, & Ntakimazi, 2018) hampers atmospheric relative humidity during that period, which generates straightforward interpretation of changes involved and, too often, evaporative‐driven cooling of near‐surface waters, and secondly methodological differences between past and recent surveys fur‐ by the reduced incoming longwave radiation reaching the lake sur‐ ther complicate data processing and interpretation. Therefore, face owing to lower cloudiness (Thiery, Martynov, et al., 2014). The studies on tropical lakes have mostly relied on paleolimnological ap‐ strength of the stratification during the wet season varies from year proaches based on various sediment archives, such as fossil diatoms to year, with notably weak stratification periods during the 2008 and (Haberyan & Hecky, 1987) or phytoplankton pigments (Saulnier‐ 2009 wet seasons. Overall, the recent years have been marked by a Talbot et al., 2014), with a time resolution of decades to centuries. slight reduction in stratification strength (Figure 2), possibly related In deep tropical lakes, phytoplankton composition responds pri‐ to the transition from a wet to a drier period. marily to the alternation of stratified conditions in the rainy season Whereas moderate to severe P‐limitation of phytoplankton pre‐ with the deep mixing that occurs in the dry season: the seasonal vails in Lake Kivu during most of the year (Sarmento, Darchambeau, mixing event not only increases the supply of dissolved nutrients & Descy, 2012; Sarmento et al., 2009), N‐limitation may occur in to the euphotic zone, but also results in a higher mixed layer (Zm): the stratified rainy season, as denitrification takes place within the euphotic layer (Zeu) depth ratio (Sarmento, Isumbisho, & Descy, oxic–anoxic transition zone (Lliros et al., 2012; Roland et al., 2018). 2006). In these conditions, phytoplankton is subjected to potential Pelagic primary production in Lake