According to the satellite remote sensing images of different years (Fig. 2), the expansion process of studied mangrove swamp from 1988 to 2015 is re-established. As shown in Figs 2a and 3, the expanding process of mangrove swamp is not gradual, the notable area expansion only took place in some special periods: from 1988 AD to 1993 AD (e1) and from 2005 AD to 2013 AD (e2). During e1, the area of mangrove increased from 60.4 hm² to 72.0 hm² with an annual increasing rate of 2.1 hm²/a. During e2, the area of mangrove increased from 75.0 hm² to 134.0 hm² with an annual increasing rate of 7.4 hm²/a. During the period between e1 and e2, the change of mangrove swamp area is very little with a increasing rate of only 0.16 hm²/a. Moreover, we can see obvious young mangrove patches (with light green color due to the light reflection of tidal flat through a sparse and small crown of young mangroves) in satellite images of e1 and e2 (for example: in 1991, 1993, 2013 and 2015). Given the fact that the mature Aegiceras corniculatum can generate a large number of seedlings every year, we can judge that the mangrove seedlings transported to bare flat only can survive during special short periods (e1 and e2) over the past 30 years.

Mangroves can survive only when their substrate elevation exceeds a low threshold value, which is usually a little higher than the local mean seal level (Zhang et al., 1997; Ye et al., 2003). The colonization of mangroves over recent decades might be attributed to the elevation lifting of bare tidal flats. To reveal the possible relation between mangrove expansion and tidal flat elevation change, the history of the mangrove substrate elevation must be reconstructed. Firstly, the minimum elevation for Aegiceras corniculatum seedling survival is determined as 0.38 m above the local mean sea level, which was obtained by measuring the substrate elevation of mangrove seedling fringe on the slope of a channel bar (Fig. 4a, Figs 5a and b). The mangrove tidal flat elevation (in 2013) was detected along the main expansion direction of the mangrove swamp (Fig. 5a). The results show that the mangrove substrate elevation decreased from 1.4 m in the oldest mangrove patch to 0.4 m in the youngest mangrove patch (Fig. 5b). A core was collected at a position just in front of the mangrove fringe of 2005. The excess ²¹⁰Pb activity decreases in a fluctuation way down core, this indicates non-negligible time-varying of accretion rate. With the CRS model, the accumulation rate and sedimentation age of each depth in core can be calculated. As is shown in Fig. 5d, the accumulation rate in surface layer (0–10 cm) is much higher than that in the underling layer. The sedimentation age is calculated to be 10 years (in 2003) at depth of 10 cm, when the mangrove fringe had just reached the position of Core C7 (Fig. 5a). The high accumulation rate in surface layer is likely caused by mangrove colonization after 2003, which can favor sedimentation by reducing hydrodynamic energy. Without taking into account the sea level rising and the compaction of underlying sediment layer in past decades, the sedimentation age is about 40 years at the minimum elevation of mangrove survival in core, which means that the tidal flat elevation had reached the low threshold no later than 1973.

Mangrove vegetation can be greatly affected by climate factors such as temperature, precipitation and storm events. For example, mangrove seedlings or even mature mangrove trees can be frozen to death if extreme low temperature events emerge in winter (Woodroffe and Grindrod, 1991; Snedaker, 1995). Precipitation in the drainage basin can affect the sediment/nutrient flux into the sea and the estuarine salinity. High sediment/nutrient flux is favorable for mangrove seedling survival by increasing the accretion rate of tidal flat and the growing speed of mangrove seedling (Nardin et al., 2016). In storm events, the mangrove seedlings in front of the mangrove fringe can be threatened both by the prolonged inundation time and by the substrate erosion caused by strong storm surges (Baldwin et al., 2001; Sherman et al., 2001).

To reveal the relation between mangrove expansion and climate change, we collected annual meteorological records of 1975 AD to 2015 AD in the study area. Time series of annual maximum wind speed (MWS) during typhoon landing (indicating the maximum intensity of landing typhoons in a year; the value is set as 0 if no typhoon landing in a year), minimum temperature (indicating winter cold events) and precipitation (indicating the salinity of estuary) were analyzed together. This is shown in Fig. 6. The years with both low typhoon intensity (MWS is less than 10 m/s, the emergency probability of MWS greater than 10 cm/s is 50%) and high annual minimum temperature (more than an average value of 3.8°C) are determined as climate optimum periods for mangrove seedling survival. From 1975 AD to 2015 AD, the total duration of climate optimum period is 14 years, which account for 34% of the whole time span. It is notable that the two mangrove expanding events (e1: 1988 AD to 1993 AD, e2: 2005 AD to 2012 AD) started almost synchronously with the two longest climate optimum periods (P1: 1988 AD to 1993 AD, P2: 2004 AD to 2008 AD). It seems a large number of mangrove seedlings can survive successfully only during continuous years of low typhoon intensity and high minimum temperature in winter. The rain fall shows no direct relation with mangrove expansion; the mangrove expansion can take place both in the relative wet (e1) and dry (e2) period. It seems that the rain fall can not affect the mangrove expansion directly in this area.

The elevation of the substrate exceeding a certain threshold is a precondition for mangrove seedling survival. According to this study, at the position of Core C7 collected, the surface elevation of tidal flat had reached the low threshold for mangrove seedling survival no later than 1973 AD. However, mangroves had not colonized there until 2003. There is a time lag of more than 30 years between the substrate elevation reaching low threshold and the emerging of mangroves. The distance from position of Core C7 to the oldest mangrove patch was about 200 m (Fig. 5a). The mangrove hypocotyls can be easily transported there by ebb current. This time lag reminds us there must be some other factors directly controlling the survival of mangrove seedling after the substrate elevation reaching the low threshold. If we only take into account the elevation of bare tidal flat in forecasting mangrove swamp expansion in time scale of decades, the result would be overestimated in this area.

The mangrove swamp in this study is located in an open estuary of Southwest China, which is in the region of northern tropic. The typhoon and cold events in winter (caused by strong East Asia Winter Monsoon) thus play very important roles in controlling mangrove seedling survival and the expansion of mangroves. According to this study, the notable mangrove expansion events can only be observed to take place synchronously with the continuous years of “climate optimum”. In the separate year of climate optimum, the mangrove seedlings may survive temporarily, but they will be eroded by energetic storm surges or killed by the cold events in the following years. For example, from 1998 AD to 2000 AD, there was no typhoon, but the low minimum temperature (3.6°C) in 1999 AD interrupt the continuous years of climate optimum. The time span of mangrove expanding events are longer than that of corresponding climate optimum period (Fig. 6), which might be caused by the time lag between the seedlings survival and their growing up enough to be identified in satellite images.

The mangrove expansion during this period thus is very little. During the past decades, continuous years of climate optimum are scarce, which caused a non-progressive expanding process of mangrove swamp (Fig. 7). Compared with typhoon intensity and annual minimum temperature, the rainfall shows no direct relation with mangrove expansion. The rainfall usually relates to mangrove expansion through affecting the accretion rate of tidal flat and the salinity of estuary water. In the study area, the substrate elevation is proven unable to control mangrove expansion directly in the time scale of decades. The rainfall of the drainage basin thus cannot control mangrove expansion by affecting the accretion rate of bare tidal flat. Moreover, the rainfall usually exceeds 1 500 mm/a in this area even in “dry” years during the past 40 years, the estuarine salinity can maintain a suitably low value for Aegiceras corniculatum seedlings.

Hydrodynamic conditions can affect the expansion of mangroves by controlling the sedimentation of mangrove seedlings. The decline of ebb current speed or wave energy thus can favor the expansion of mangroves. In the study area, most of the wave energy is dissipated by bottom friction due to the wide tidal flat (5–10 km in width) in condition of normal weather^①. The wave disturbance is very weak in front of mangrove swamp and can not affect the sedimentation of mangrove seedlings significantly. The seedlings can settle down and survive on the channel bar slope, which habitat elevation is 0.38–0.63 m above mean sea level. Considering the fact that the ebb current speed on tidal flat is lower than that on the slope of channel bar with same elevation, the seedlings must be able to settle down and survive temporally on the tidal flat where Core C7 is collected no later than 1973 (Fig. 5e). However, the mangrove cannot colonized there until 2003. It seems that the decline of ebb current speed due to tidal flat accretion is not a key factor affecting mangrove swamp expansion in this area.

The N and P nutrient flux transported into estuary might also be a factor affecting mangrove swamp expansion. The N and P nutrient flux into the Nanliu River Estuary increased sharply after the 1980s due to the rapid development of local agriculture and aquaculture (Xia et al., 2011). Increased nutrients in the sediments might promote the growth and survival of mangrove seedlings (Koch and Snedaker, 1997), but cannot explain the discontinuity of the mangrove swamp expanding process either. Nutrients supply thus is not an important factor affecting mangrove swamp expansion in this area.

The results of this study can help us predict the future fate of studied the mangrove swamp in consideration of future climate change and sea level rise. Due to global warming, the emerging frequency of energy cyclones is very likely to increase in the future (Emanuel, 2005). This climate change trend also can be observed in the study area (Fig. 6), which shows an overall increasing of the annual minimum temperature from the 1970s and a notable typhoon intensity increasing after 2009. If the landing frequency of powerful typhoons increases in the future, the climate optimum period for mangrove seedling survival will be more scarce, which will limit the further expansion of mangrove swamps. Moreover, more energetic storm surges can restrain the elevation lifting of bare tidal flat by their compressing and eroding effect (Van Santen et al., 2007), which also obstructs the mangrove swamp from expanding seaward. For example, the accretion rate of bare flat before the studied mangrove swamp is 3–5 mm/a (Fig. 5d) in the past decades, which will decease if the frequency of powerful typhoons increases in future. Considering the global sea level rise will be 28–98 cm at the end of 21centry (IPCC, 2013), the elevation of bare tidal flats will very likely be unable to keep up with the sea level rise, and the area of possible habitat for mangrove swamp will be reduced.